Composition of the Continental Crust

4.1

Composition of the Continental Crust

RL Rudnick, University of Maryland, College Park, MD, USA S Gao, China University of Geosciences, Wuhan, People’s Republic of China and Northwest University, Xi’an, People’s Republic of China ã 2014 Elsevier Ltd. All rights reserved. This article is a revision of the previous edition article, volume 3, pp. 1–64, © 2003, Elsevier Ltd.

4.1.1 Introduction 4.1.1.1 What is the Continental Crust? 4.1.1.2 The Importance of Determining Crust Composition 4.1.2 The Upper Continental Crust 4.1.2.1 Surface Averages 4.1.2.2 Sedimentary Rocks and Glacial Deposit Averages 4.1.2.2.1 Sedimentary rocks 4.1.2.2.2 Glacial deposits and loess 4.1.2.3 An Average Upper-Crustal Composition 4.1.3 The Deep Crust 4.1.3.1 Definitions 4.1.3.2 Metamorphism and Lithologies 4.1.3.3 Methodology 4.1.3.4 The Middle Crust 4.1.3.4.1 Samples 4.1.3.4.2 Seismological evidence 4.1.3.4.3 Middle-crust composition 4.1.3.5 The Lower Crust 4.1.3.5.1 Samples 4.1.3.5.2 Seismological evidence 4.1.3.5.3 Lower-crust composition 4.1.4 Bulk Crust Composition 4.1.4.1 A New Estimate of Crust Composition 4.1.4.2 Intracrustal Differentiation 4.1.5 Implications of the Crust Composition 4.1.6 Earth’s Crust in a Planetary Perspective 4.1.7 Summary Acknowledgments References

4.1.1

Introduction

The Earth is an unusual planet in our solar system in having a bimodal topography that reflects the two distinct types of crust found on our planet. The low-lying oceanic crust is thin (
7 km on average), composed of relatively dense rock types such as basalt and is young (200 Ma old) (see Chapter 4.13 and 4.14). In contrast, the high-standing continental crust is thick (
40 km on average), is composed of highly diverse lithologies (virtually every rock type known on Earth) that yield an average intermediate or ‘andesitic’ bulk composition (Taylor and McLennan (1985) and references therein), and contains the oldest rocks and minerals yet observed on Earth (currently the 4.0 Ga Acasta gneisses (Bowring and Williams, 1999) and 4.4 Ga detrital zircons from the Yilgarn Block, Western Australia (Wilde et al., 2001)), respectively. Thus, the continents preserve a rich geological history of our planet’s evolution and understanding their origin is critical for understanding the origin and differentiation of the Earth.

Treatise on Geochemistry 2nd Edition

1 2 3 3 3 9 10 12 16 17 18 18 18 19 19 22 22 23 23 31 32 36 41 41 42 44 45 45 45

The origin of the continents has received wide attention within the geological community, with hundreds of papers and several books devoted to the topic (the reader is referred to the following general references for further reading: Taylor and McLennan (1985), Windley (1995), and Condie (1997). Knowledge of the age and composition of the continental crust is essential for understanding its origin. Patchett and Samson (2003) review the present-day age distribution of the continental crust and Kemp and Hawkesworth Chapter 4.11 review secular evolution of crust composition. Moreover, to understand fully the origin and evolution of continents requires an understanding of not only the crust, but also the mantle lithosphere that formed more-or-less contemporaneously with the crust and translates with it as the continents move across the Earth’s surface. The latter topic is reviewed in Chapters 3.6 and 3.8. This chapter reviews the present-day composition of the continental crust, the methods employed to derive these estimates, and the implications of the continental crust

http://dx.doi.org/10.1016/B978-0-08-095975-7.00301-6

1

2

Composition of the Continental Crust

Figure 1 Map of continental regions of the Earth, including submerged continents. (Cogley (1984); reproduced by permission of American Geophysical Union from Reviews of Geophysics and Space Physics, 1984, 22, 101122).

composition for the formation of the continents, Earth differentiation, and its geochemical inventories.

4.1.1.1

What is the Continental Crust?

In a review of the composition of the continental crust, it is useful to begin by defining the region under consideration and to provide some generalities regarding its structure. The continental crust, as considered here, extends vertically from the Earth’s surface to the Mohorovicic discontinuity, a jump in compressional wave speeds from
7 km s1 to
8 km s1 that is interpreted to mark the crust–mantle boundary. In some regions the Moho is transitional rather than discontinuous and there may be some debate as to where the crust–mantle boundary lies (cf. Griffin and O’Reilly, 1987; McDonough et al., 1991). The lateral extent of the continents is marked by the break in slope on the continental shelf. Using this definition,
31% of continental area is submerged beneath the oceans (Figure 1; Cogley, 1984), and is thus less accessible to geological sampling. For this reason, most estimates of continental crust composition derive from exposed regions of the continents. In some cases the limited geophysical data for submerged continental shelves reveal no systematic difference in bulk properties between the shelves and exposed continents; the shelves simply appear to be thinned regions of the crust. In other cases, such as volcanic rifted margins, the submerged continent is characterized by high-velocity layers interpreted to represent massive basaltic intrusions associated with continental breakup (Holbrook and Kelemen, 1993). Depending on the extent of the latter type of continental margin (which is yet to be quantified), crust compositional estimates derived from exposed regions may not be wholly representative of the total continental mass. The structure of the continental crust is defined seismically to consist of upper-, middle-, and lower crustal layers (Christensen and Mooney, 1995; Holbrook et al., 1992; Rudnick and Fountain, 1995). The upper crust is readily

accessible to sampling and robust estimates of its composition are available for most elements (Section 4.1.2). These show the upper crust to have a granodioritic bulk composition, to be rich in incompatible elements, and generally depleted in compatible elements. The deeper reaches of the crust are more difficult to study. In general, three probes of the deep crust are employed to unravel its composition: (1) studies of highgrade metamorphic rocks (amphibolite or granulite facies) exposed in surface outcrops (Bohlen and Mezger, 1989) and, in some cases, in uplifted cross-sections of the crust reaching to depths of 20 km or more (Fountain et al., 1990a; Hart et al., 1990; Ketcham, 1996; Miller and Christensen, 1994); (2) studies of granulite-facies xenoliths (foreign rock fragments) that are carried from great depths to the Earth’s surface by fast-rising magmas (see Rudnick (1992) and references therein); and (3) remote sensing of lower crustal lithologies through seismic investigations (Christensen and Mooney, 1995; Holbrook et al., 1992; Rudnick and Fountain, 1995; Smithson, 1978) and surface heat-flow studies (see Chapter 4.2). Collectively, the observations from these probes show that the crust becomes more mafic with depth (Section 4.1.3). In addition, the concentration of heat-producing elements drops off rapidly from the surface downwards. This is due, in part, to an increase in metamorphic grade but is also due to increasing proportions of mafic lithologies (see Chapter 4.2). Thus, the crust is vertically stratified in terms of its chemical composition. In addition to this stratification, the above studies also show that the crust is heterogeneous from place to place, with few systematics available for making generalizations about crustal structure and composition for different tectonic settings. For example, the crust of Archean cratons in some regions is relatively thin and has low seismic velocities, suggesting an evolved composition (e.g., Yilgarn craton (Drummond, 1988); Kaapvaal craton (Durrheim and Green, 1992; Niu and James, 2002); and North China craton (Gao et al., 1998a,b)). However, in other cratons, the crust is thick (4050 km)

Composition of the Continental Crust

and the deep crust is characterized by high velocities, which imply mafic bulk compositions (Wyoming craton (Gorman et al., 2002) and Baltic shield (Luosto and Korhonen, 1986; Luosto et al., 1990)). The reasons for these heterogeneities are not fully understood and we return to this topic in Section 4.1.3. Similar heterogeneities are observed for Proterozoic and Paleozoic regions (see Rudnick and Fountain (1995) and references therein). Determining an average composition of such a heterogeneous mass is difficult and, at first glance, may seem like a futile endeavor. Yet it is just such averages that allow insights into the relative contribution of the crust to the whole Earth-chemical budget and the origin of the continents. Thus, deriving average compositions is critical to studies of the continents and the whole Earth.

4.1.1.2

The Importance of Determining Crust Composition

Although the continental crust constitutes only
0.6% by mass of the silicate Earth, it contains a very large proportion of incompatible elements (20–70%, depending on element and model considered; Rudnick and Fountain (1995)), which include the heat-producing elements and members of a number of radiogenic-isotope systems (RbSr, UPb, SmNd, LuHf). Thus the continental crust factors prominently in any mass-balance calculation for the Earth as a whole and in estimates of the thermal structure of the Earth (Sclater et al., 1980). In addition, knowledge of the bulk composition of the crust and determining whether this composition has changed through time is important for: (1) understanding the processes by which the crust is generated and modified and (2) determining whether there is any secular evolution in crust generation and modification processes (see Chapter 4.11). The latter has important implications for the evolution of our planet as a whole. In this chapter we review the composition of the upper, middle, and lower continental crust (Sections 4.1.2 and 4.1.3). We then examine the bulk crust composition and the implications of this composition for crust generation and modification processes (Sections 4.1.4 and 4.1.5). Finally, we compare the Earth’s crust with those of the other terrestrial planets in our solar system (Section 4.1.6) and speculate about what unique processes on Earth have given rise to this unusual crustal distribution.

4.1.2

The Upper Continental Crust

The upper continental crust, being the most accessible part of our planet, has long been the target of geochemical investigations (Clarke, 1889). There are two basic methods employed to determine the composition of the upper crust: (1) establishing weighted averages of the compositions of rocks exposed at the surface and (2) determining averages of the composition of insoluble elements in fine-grained clastic sedimentary rocks or glacial deposits and using these to infer upper-crust composition. The first method was utilized by F. W. Clarke and colleagues over a century ago (Clarke, 1889; Clarke and Washington, 1924) and entails large-scale sampling and weighted averaging

3

of the wide variety of rocks that crop out at the Earth’s surface. All major-element (and a number of soluble trace elements) determinations of upper-crust composition rely upon this method. The latter method is based on the concept that the process of sedimentation averages wide areas of exposed crust. This method was originally employed by Goldschmidt (1933) and his Norwegian colleagues in their analyses of glacial sediments to derive average composition of the crystalline rocks of the Baltic shield and has subsequently been applied by a number of investigators, including the widely cited work by Taylor and McLennan (1985) to derive upper-crust composition for insoluble trace elements. In the following sections we review the upper-crust composition determined from each of these methods, then provide an updated estimate of the composition of the upper crust.

4.1.2.1

Surface Averages

In every model for the composition of the upper-continental crust, major-element data are derived from averages of the composition of surface exposures (Table 1). Several surfaceexposure studies have also provided estimates of the average composition of a number of trace elements (Table 2). For soluble elements that are fractionated during the weathering process (e.g., sodium, calcium, strontium, barium, etc.), this is the only way in which a reliable estimate of their abundances can be obtained. The earliest of such studies was the pioneering work of Clarke (1889), who, averaging hundreds of analyses of exposed rocks, determined an average composition for the crust that is markedly similar to present-day averages of the bulk crust (cf. Tables 1 and 9). Although Clarke’s intention was to derive the average crust composition, his samples are limited to the upper crust; there was little knowledge of the structure of the Earth when these studies were undertaken; oceanic crust was not distinguished as different from continental and the crust was assumed to be only 16 km thick. Clarke’s values are, therefore, most appropriately compared to upper crustal estimates. Later, Clarke, joined by H. S. Washington, used a larger data set to determine an average composition of the uppercrust that is only slightly different from his original 1889 average (Clarke and Washington, 1924; Table 1). Compared to more recent estimates of upper-crust composition, these earliest estimates are less evolved (lower silicon, higher iron, magnesium, and calcium), but contain similar amounts of the alkali elements, potassium and sodium. The next major undertakings in determining upper-crust composition from large-scale surface sampling campaigns did not appear until twenty years later in studies centered on the Canadian, Baltic, and Ukranian Shields. It is these studies that form the foundation on which many of the more recent estimates of upper-crust composition are constructed (e.g., Taylor and McLennan, 1985; Wedepohl, 1995). Shaw et al. (1967, 1976, 1986) and Eade and Fahrig (1971, 1973) independently derived estimates for the average composition of the Canadian Precambrian shield. Both studies created composites from representative samples taken over large areas that were weighted to reflect their surface outcrop area. The estimates of Shaw et al. are based on a significantly smaller

4

Element

1 Clarke (1889)

2 Clarke and Washington (1924)

3 Goldschmidt (1933)

4 Shaw et al. (1967)

5 Fahrig and Eade (1968)

6 Ronov and Yaroshevskiy (1976)

7 Condie (1993)

8 Gao et al. (1998a)b

9 Borodin (1998)

10 Taylor and McLennan (1985)

11 Wedepohl (1995)

12 This Study c

SiO2 TiO2 Al2O3 FeOdT MnO MgO CaO Na2O K2O P2O5 Mg#

60.2 0.57 15.27 7.26 0.10 4.59 5.45 3.29 2.99 0.23 53.0

60.30 1.07 15.65 6.70 0.12 3.56 5.18 3.92 3.19 0.31 48.7

62.22 0.83 16.63 6.99 0.12 3.47 3.23 2.15 4.13 0.23 46.9

66.8 0.54 15.05 4.09 0.07 2.30 4.24 3.56 3.19 0.15 50.1

66.2 0.54 16.10 4.40 0.08 2.20 3.40 3.90 2.91 0.16 47.4

64.8 0.55 15.84 5.78 0.10 3.01 3.91 2.81 3.01 0.16 48.1

67.0 0.56 15.14 4.76

67.97 0.67 14.17 5.33 0.10 2.62 3.44 2.86 2.68 0.16 46.7

67.12 0.60 15.53 4.94 0.00 2.10 3.51 3.21 3.01 0.00 43.2

65.89 0.50 15.17 4.49 0.07 2.20 4.19 3.89 3.39 0.20 46.6

66.8 0.54 15.05 4.09 0.07 2.30 4.24 3.56 3.19 0.15 50.1

66.62 0.64 15.40 5.04 0.10 2.48 3.59 3.27 2.80 0.15 46.7

a

Major elements recast to 100% anhydrous. Carbonate-free estimate. c See Table 3 for derviation of this estimate. d Total Fe as FeO. Mg# ¼ molar 100  Mg/(Mg þ Fetot). b

2.45 3.64 3.55 2.76 0.12 47.9

Composition of the Continental Crust

Table 1 Major element compositiona (in weight percent oxide) of the upper continental crust. Columns 19 represent averages of surface exposures and glacial clays. Columns 1011 are derivative compositions from these data. Column 12 shows our recommended values

Table 2 Estimates of the trace-element composition of the upper continental crust. Columns 1–4 represent averages of surface exposures. Columns 5–8 are estimates derived from sedimentary and loess data. Column 9 is a previous estimate, where bracketed data are values derived from surface exposure studies. Column 10 is our recommended value (see Table 3)

Li Be B N F S Cl Sc V Cr Co Ni Cu Zn Ga Ge As Se Br Rb Sr Y Zr Nb Mo Ru Pd Ag Cd In Sn Sb I Cs Ba La Ce Pr

mg g 1 ” ” ” ” ” ” ” ” ” ” ” ” ” ” ” ” ” ” ” ” ” ” ” ” ng g1 ” ” mg g1 ” ” ” ” ” ” ” ” ”

1 Shaw et al. (1967, 1976)

2 Eade and Fahrig (1973)

3 Condie (1993)

22 1.3 9.2 500 600 100 7 53 35 12 19 14 52 14

110 316 21 237 26

12 59 76 19 26 60

85 380 21 190

13.4 86 112 18 60

83 289 24 160 9.8

4 Gao et al. (1998a)a

6 Plank and Langmuir (1998)

633 28.4 57.5

8 Taylor and McLennan (1985, 1995) 20 3 15

561 309 142 15 98 80 17 38 32 70 18 1.34 4.4 0.15

13.6d 107d 85d 17d 44d 25 71 17 1.6 1.5 0.05

82 266 17.4 188 12 0.78

1.73 0.3

730 71

7 Peucker-Eherenbrink and Jahn (2001)

20 1.95 28

5.1

3.55 678 34.8 66.4

112 350 22 190 12d 1.5

13.7 1.2 0.34 0.52

1.46 55 0.079

0.075

1070 32.3 65.6

5 Sims et al. (1990)

0.45 7.3

0.5 50 0.098 0.05 5.5 0.2 4.6d 550 30 64 7.1

9 Wedepohl (1995)b

10 This Studyc

[22] 3.1 17 83 611 953 640 [7] [53] [35] [12] [19] [14] [52] [14] 1.4 2 0.083 1.6 110 [316] [21] [237] [26] 1.4

21 2.1 17 83 557 621 370 14.0 97 92 17.3 47 28 67 17.5 1.4 4.8 0.09 1.6 84 320 21 193 12 1.1 0.34 0.52 53 0.09 0.056 2.1 0.4 1.4 4.9 624 31 63 7.1 (Continued)

5

Units

Composition of the Continental Crust

Element

55 0.102 0.061 2.5 0.31 1.4 5.8 668 [32.3] [65.7] 6.3

(Continued)

6

Table 2

Units

1 Shaw et al. (1967, 1976)

Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu Hf Ta W Re Os Ir Pt Au Hg T1 Pb Bi Th U

” ” ” ” ” ” ” ” ” ” ” ” ” ” ng g–1 ” ” ” ” mg g–1 ” ” ” ” ”

25.9 4.61 0.937

a

2 Eade and Fahrig (1973)

0.481 2.9 0.62

1.47 0.233 5.8 5.7

3 Condie (1993)

4 Gao et al. (1998a)a

25.6 4.59 1.05 4.21 0.66

30.4 5.09 1.21

1.91 0.32 4.3 0.79

2.26 0.35 5.12 0.74 0.91

6 Plank and Langmuir (1998)

7 Peucker-Eherenbrink and Jahn (2001)

0.82

0.96 3.3 0.198 0.031 0.022 0.51

0.02 1.81 0.096 0.524 17 0.035 10.3 2.45

5 Sims et al. (1990)

18 10.8 1.5

17 8.6 2.2

1.24 0.0123 1.55 18 0.23 8.95 1.55

Carbonate-free estimate. Wedepohl’s upper crust is largely derived from the Canadian Shield composites of Shaw et al. (1967, 1976). Values taken directly from Shaw et al. are shown in brackets. c See Table 3 for derviation of this estimate. d Updated in McLennan (2001b). b

8 Taylor and McLennan (1985, 1995) 26 4.5 0.88 3.8 0.64 3.5 0.8 2.3 0.33 2.2 0.32 5.8 1.0d 2 0.4 0.05 [0.02]

9 Wedepohl (1995)b

4.7 0.95 2.8 [0.5] [2.9] [0.62]

[1.5] [0.27] [5.8] 1.5 1.4

[1.8] 0.75 17d 0.13 10.7 2.8

0.056 0.75 17 0.123 [10.3] [2.5]

10 This Studyc

27 4.7 1.0 4.0 0.7 3.9 0.83 2.3 0.30 2.0 0.31 5.3 0.9 1.9 0.198 0.031 0.022 0.5 1.5 0.05 0.9 17 0.16 10.5 2.7

Composition of the Continental Crust

Element

Composition of the Continental Crust

Normalized to UCR and G

1.4 1.2

7

Shaw et al. Eade and Fahrig Taylor and McLennan

1 0.8 0.6

(a)

Si

Al

Fe

Mg

Ca

Na

K

Mg

Ca

Na

K

Normalized to UCR and G

1.4 1.2 1 0.8 0.6 (b)

Borodin Condie Gao et al. Ronov and Yaroshevsky

Si

Al

Fe

Figure 2 Comparison of different models for the major-element composition of the upper continental crust. All values normalized to the new composition provided in Table 3. Gray shaded field represents 10% variation from this value. (a) Compositions derived from Canadian Shield samples (Shaw et al., 1967, 1976, 1986; Fahrig and Eade, 1968; Eade and Fahrig, 1971, 1973) and the Taylor and McLennan model (1985, 1995, as modified by McLennan, 2001b). (b) Compositions derived from surface sampling of the former Soviet Union (Ronov and Yaroshevsky, 1967, 1976; Borodin, 1998) and China (Gao et al., 1998a) and a global compilation of upper crustal rock types weighted in proportion to their areal distribution (Condie, 1993). The Canadian shield averages appear to be more evolved (having lower Mg, Fe, and higher Na and K) than other estimates.

number of samples than that of Eade and Fahrig’s (i.e.,
430 versus
14 000) and cover different regions of the shield, but the results are remarkably similar (Figure 2). All major elements agree to within
10% except for CaO, which is
20% higher, and MnO, which is 15% lower in the estimates of Shaw et al. Shaw et al. (1967, 1976, 1986) also measured a number of trace elements in their shield composites and these are compared to the smaller number of trace elements determined by Eade and Fahrig in Table 2 and Figure 3. As might be expected, considering the generally greater variability in trace-element concentrations and the greater analytical challenge, larger discrepancies exist between the two averages. For example, scandium, chromium, copper, lanthanum, and uranium values vary by
50% or more. In some cases this may reflect compromised data quality (e.g., lanthanum was determined by optical-emission spectroscopy in the Eade and Fahrig study) and in other cases it may reflect real differences in the composition between the two averages. However, for a number of trace elements (e.g., vanadium, nickel, zinc, rubidium, strontium, zirconium, and thorium), the averages agree within 30%. In a similar study, Ronov and Yaroshevsky (1967, 1976) determined the average major-element composition of the upper crust based on extensive sampling of rocks from the Baltic and Ukranian shields and the basement of the Russian platform (Table 1). While the SiO2, Al2O3, and K2O values fall within 5% of those of the average Canadian Shield, as

determined by Eade and Fahrig (1971, 1973), FeOT, MgO, and CaO are
1030% higher, and Na2O is
30% lower than the Canadian average, suggesting a slightly more mafic composition. The generally good correspondence between these independent estimates of the composition of shield upper crust lends confidence in the methodologies employed. However, questions can be raised about how representative the shields are of the global upper continental crust. For example, Condie (1993) suggests that shield averages may be biased because (1) shields are significantly eroded and thus may not be representative of the 5–20 km of uppermost crust that has been removed from them and (2) they include only Precambrian upper crust and largely ignore any Phanerozoic contribution to upper crust. Condie (1993) derived an upper-crust composition based on over 3000 analyses of upper crustal rock types weighted according to their distributions on geologic maps and stratigraphic sections, mainly covering regions of North America, Europe, and Australia. He utilized two methods in calculating an average upper-crust composition: (1) using the map distributions, irrespective of level of erosion and (2) for areas that have been significantly eroded, restoring the eroded upper crust, assuming it has a ratio of supracrustal rocks to plutonic rocks similar to that seen in uneroded upper crustal regions. The latter approach was particularly important for his study, as one of his primary objectives was to evaluate whether there has been any secular change in upper crust composition. However, in this review, we are interested in the present-day

8

Composition of the Continental Crust

2.0

2.0 Transition metals

1.5

1.5

1.0

1.0

0.5

0.5

2.2

6.5

High field-strength (a)

Mn

Sc

V

Cr

Co

Ni

Cu

Zn

2.0

(b)

Ti

Zr

Mo

W

Rare earth

1.5

1.5

1.0

1.0

0.50

0.5

0.0

Ta

2.0 Alkali and alkaline earth

(c)

Nb

Hf

Li

Rb

Cs

Be

Sr

Ba

(d)

Shaw Eade and Fahrig Condie

La

Ce

Nd Sm Eu

Taylor and McLennan Gao et al.

Gd

Tb

Yb

Lu

10

2.0 Actinides and heavy metals

Siderophile and chalcophile

1.5 1.0

1.0 0.5

Log scale 0.10 (e)

Tl

Pb

Bi

Th

U

(f)

Pd

Ag

Cd

In

Sn

Sb

Au

Hg

Figure 3 Comparison of different models for the trace-element composition of the upper-continental crust. All values normalized to the new composition provided in Table 3. Gray shaded field represents 20% variation from this value for all panels except (f), in which gray field represents a factor of two variation. Trace elements are divided into the following groups: (a) transition metals, (b) high-field strength elements, (c) alkali, alkaline-earth elements, (d) REEs, (e) actinides and heavy metals, and (f) highly siderophile and chalcophile elements (note log scale). Data from Tables 1 and 2; lanthanum estimate from Eade and Fahrig (1973) is omitted from panel D.

composition of the upper crust (eroded or not), so it seems most appropriate to consider his ‘map model’ for comparisons with other models (for a discussion of the secular evolution of the continents, see Chapter 4.11). Condie’s ‘map model’ is compared with other estimates of the upper crust in Tables 1 and 2 and Figures 2 and 3. For major elements, his upper crust composition is within 10% of the Canadian Shield values of Eade and Fahrig. It is also within 10% of some of the major elements estimated by Shaw, but has generally higher magnesium and iron, and lower calcium and potassium compared to Shaw’s estimate (Figure 2). Many trace elements in Condie’s upper-crust composition are similar (i.e., within 20%) to those of Shaw’s Canadian Shield composites (Figure 3), including the light rare-earth elements (LREEs), strontium, yttrium, thorium, and uranium. However, several trace elements in Condie’s average vary by 50% from those of Shaw et al. (1967, 1976, 1986) as can be seen in the figure. These include transition metals (scandium, vanadium, chromium, and nickel), which are considerably higher in Condie’s upper crust, and niobium, barium and tantalum, which are

significantly lower in Condie’s upper crust compared to Shaw’s. These differences may reflect regional variations in upper crust composition (i.e., the Canadian Shield is not representative of the worldwide upper crust) or inaccuracies in either of the estimates due to data quality or insufficiency. As will be discussed below, it is likely that Condie’s values for transition metals, niobium, tantalum, and barium are the more robust estimates of the average upper crust composition. A recent paper by Borodin (1998) provides an average composition of the upper crust that includes much Soviet shield and granite data not included in most other worldwide averages. For this reason, it makes an interesting comparison with other data sets. Like other upper crustal estimates, major elements in the Borodin average upper crust (Table 1 and Figure 2) fall within 10% of the Eade and Fahrig average for the Canadian Shield, except for TiO2 and FeO, which are
13% higher, and Na2O, which is
20% lower than the Canadian average. Borodin’s limited trace element averages (for chromium, nickel, rubidium, strontium, zirconium, niobium, barium, lanthanum, thorium, and uranium – not given

Composition of the Continental Crust

in table or figures) fall within 50% of Shaw’s Canadian Shield values except for niobium, which, like other upper crustal estimates, is about a factor of 2 lower than the Canadian average. The most recent and comprehensive study of upper-crust composition derived from surface exposures was carried out by Gao et al. (1998a). Nine hundred and five composite samples were produced from over 11 000 individual rock samples covering an area of 9.5  105 km2 in eastern China, which includes samples from Precambrian cratons as well as Phanerozoic fold belts. The samples comprised both crystalline basement rocks and sedimentary cover, the thickness of which was determined from seismic and aeromagnetic data. Averages were derived by combining compositions of individual map units weighted according to their thicknesses (in the cases of sedimentary cover) and areal exposure, for shields. The upper crust is estimated to be
15 km thick based on seismic studies (Gao et al., 1998a) and the crystalline rocks exposed at the surface are assumed to maintain their relative abundance through this depth interval. Average upper crust was calculated both as a grand average and on a carbonate-free basis; carbonates comprise a significant rock type (722%) in many of the areas sampled (e.g., Yangtze craton). The grand average (including carbonate) has a significantly different bulk composition than other estimates of the upper crust (Gao et al., 1998a; Table 2). Most of the latter are derived from crystalline shields and so a difference is expected. However, Condie’s map model incorporates sedimentary cover as well as crystalline basement. The differences between Condie’s map model and Gao et al. grand-total upper crust suggest that the carbonate cover in eastern China is thicker than most other areas. For this reason, we use Gao et al. (1998a) carbonate-free compositions in further discussions, but with the caveat that carbonates may be an overlooked upper crustal component in many upper crustal estimates. The Gao et al. (1998a) major- and trace-element results are presented in Tables 1 and 2 and plotted in Figures 2 and 3, respectively. Unlike the model of Condie (1993), several of the major elements fall beyond 10% of Eade and Fahrig’s Canadian Shield data (Figures 2 and 3). These include TiO2, FeO, MnO, and MgO, which are higher, and Na2O, which is lower in the eastern China upper crust compared to the Canadian Shield. Gao et al. (1998a) attribute these differences to erosional differences between the two areas. Whereas the Canadian Shield composites comprised mainly metamorphic rocks of the amphibolite facies, the eastern China composites contain large proportions of unmetamorphosed supracrustal units that are considered to have, on average, higher proportions of mafic volcanics. In this respect, the Gao et al. model composition compares favorably to Condie’s map model and the Russian estimates for all major elements. However, the Na2O content of the eastern China upper crust is one of lowest of all (
20% lower than Condie’s average and 10% lower than Borodin’s values, but similar to Ronov and Yaroshevsky’s average) (Figure 2). The trace-element composition estimated by Gao et al. (1998a) for the Chinese upper crust is very similar to that of Condie (1993). Like the latter model, many lithophile trace elements in the Gao et al. model are within 50% of the Canadian Shield averages of Shaw et al. (e.g., LREEs, yttrium,

9

rubidium, strontium, zirconium, hafnium, thorium, and uranium), and the Chinese average has significantly higher transition metals and lower niobium, barium, and tantalum than the Canadian Shield average. In addition, Gao et al. (1998a) provide values for some of the less well-constrained element concentrations. Of these, averages for lithium, beryllium, zinc, gallium, cadmium, and gold fall within 40% of the Shaw et al. averages, but boron, thallium, and bismuth are significantly higher, and mercury is significantly lower in Gao’s average than in Shaw’s. There is too little information for these elements in general to fully evaluate the significance of these differences. Two generalizations can be made from the above studies of surface composites. 1. Major element data are very consistent from study to study, with most major-element averages falling within 10% of Eade and Fahrig’s Canadian Shield average. When differences do occur, they appear to reflect a lower percentage of mafic lithologies in the Canadian averages: all other estimates (including the Russian shield data) have higher FeO and TiO2 than the Canadian averages and most also have higher CaO and MgO (Figures 2 and 3). The Eade and Fahrig average also has higher Na2O than all other estimates (including Shaw’s estimate for the Canadian Shield). 2. Trace elements show more variation than major elements from study to study, but some lithophile trace elements are relatively constant: rare earth elements (REEs), yttrium, lithium, rubidium, caesium, strontium, zirconium, hafnium, lead, thorium, and uranium do not vary beyond 50% between studies. Transition metals (scandium, cobalt, nickel, chromium, and vanadium) are consistently lower in the Canadian Shield estimates than in other studies, which may also be attributed to a lower percentage of mafic lithologies in the Canadian Shield (a conclusion supported by studies of sediment composition, as discussed in the next section). Barium is
40% higher in the Shaw et al. average than in all other averages, including that of Eade and Fahrig, suggesting that this value is too high. Finally, niobium and tantalum are both about a factor of 2 higher in the Shaw et al. average than in any other average, suggesting that the former is not representative of the upper continental crust, a conclusion reached independently by Plank and Langmuir (1998) and Gallet et al. (1998) based on the composition of marine sediments and loess (see next section).

4.1.2.2

Sedimentary Rocks and Glacial Deposit Averages

While the large-scale sampling campaigns outlined above are the primary means by which the major-element composition of the upper continental crust has been determined, many estimates of the trace-element composition of the upper crust rely on the natural wide-scale sampling processes of sedimentation and glaciation. These methods are used primarily for elements that are insoluble during weathering and are, therefore, transported quantitatively from the site of weathering/ glacial erosion to deposition. This methodology has been especially useful for determining the REE composition of the upper crust (see Taylor and McLennan (1985) and references therein). The averages derived from each of these natural largescale samples are discussed in turn. When the upper crustal

10

Composition of the Continental Crust

that the REE patterns of shales reflect that of the average uppercontinental crust (Taylor and McLennan (1985) and references therein). Thus, Taylor and McLennan’s (1985) upper crustal REE pattern is parallel to average shale, but lower in absolute abundances due to the presence of sediments with lower REE abundances such as sandstones, carbonates, and evaporites. Using a mass balance based on the proportions of different types of sedimentary rocks, they derive an upper crustal REE content that is 80% that of post-Archean average shale. Comparison of various upper crustal REE patterns is provided in Figure 5. All estimates, whether from shales, marine sediments, or surface sampling, agree to within 20% for the LREEs and
50% for the heavy rare-earth elements (HREEs). The estimate of Shaw et al. (1976) has the lowest HREEs and if these data are excluded, the HREEs agree to within 15% between the models of Condie (1993), Gao et al. (1998a), and Taylor and McLennan (1985). Thus, the REE content of the upper continental crust is established to within 10–25%, similar to the uncertainties associated with its major-element composition. Once the REE concentration of the upper crust has been established, values for other insoluble elements can be determined from their ratios with an REE. Using the constant ratios of La / Th and La/Sc observed in shales, McLennan et al. (1980) and Taylor and McLennan (1985) estimated the upper crustal thorium and scandium contents at 11 ppm and 10.7 ppm, respectively. The scandium value increased slightly (to 13.7 ppm) and the thorium value remained unchanged when

concentration of elements is discussed, the element name is printed in italic text so that the reader can quickly scan the text to the element of interest.

4.1.2.2.1 Sedimentary rocks Processes that produce sedimentary rocks include weathering, erosion, transportation, deposition, and diagenesis. Elemental fractionation during weathering is discussed in detail by Taylor and McLennan (1985) (see also Chapters 7.1 and 9.1) and the interested reader is referred to these works for more extensive information. Briefly, elements with high solubilities in natural waters (Figure 4) have greater potential for being fractionated during sedimentary processing; thus, their concentration in fine-grained sedimentary rocks may not be representative of their source region. These elements include the alkali and alkaline-earth elements as well as boron, rhenium, molybdenum, gold, and uranium. In contrast, a number of elements have very low solubilities in waters. Their concentrations in sedimentary rocks may, therefore, provide robust estimates of the average composition of their source regions (i.e., average upper-continental crust). Taylor and McLennan (1985) identified the REEs, yttrium, scandium, thorium, and possibly cobalt as being suitably insoluble and thus providing useful information on upper crust composition. The REE patterns for post-Archean shales show striking similarity worldwide (Figure 5): they are light REE enriched, with a negative europium anomaly and relatively flat heavy REEs. This remarkable consistency has led to the suggestion

10.0

8.0

Soluble Moderately soluble Insoluble

Na Mg K U

6.0

Au

Log t

Ge CrTl

Ta Hf Nb Sn Be

2.0

Sc Th

0.0

-2.0 -10.0

Al

Y Pb

Ag

V Ba

Cd Bi

Ni Ga

Zr

As

Cs

Si

4.0

Ca Mo

Sb W

Sr Li

Se

Rb

B

Re

In Zn Cu

Ti Mn

Co REE

Fe

-8.0

-6.0

-4.0

-2.0

0.0

Log Kysw

Figure 4 Plot of residence time (expressed as log t) against seawater upper crust partition coefficient (expressed as KYsw ) (source Taylor and McLennan, 1985).

Composition of the Continental Crust

1000 PAAS NASC ES ECPAS

Chondrite normalized

Shales and loess

100

10 Loess

1 (a)

La Ce Pr Nd

Sm Eu Gd Tb Dy Ho Er Tm Yb Lu

1000

Chondrite normalized

Upper crust

100

Shaw et al. Condie, Map Gao et al. Taylor and McLennan Wedepohl This study

10

1 (b)

La Ce Pr Nd

Sm Eu Gd Tb Dy Ho Er Tm Yb Lu

Figure 5 Comparison of REE patterns between (a) average postArchean shales and loess and (b) various estimates of the upper continental crust composition. PAAS ¼ post-Archean Australian Shale (Taylor and McLennan, 1985); NASC ¼ North American shale composite (Haskin et al., 1966); ES ¼ European shale composite (Haskin and Haskin, 1966); ECPAS ¼ Eastern China post-Archean shale (Gao et al., 1998a). The loess range includes samples from China, Spitsbergen, Argentina, and France (Gallet et al., 1998; Jahn et al., 2001). Chondrite values are from Taylor and McLennan (1985).

a more comprehensive sediment data set was employed by McLennan (2001b). The sediment-derived scandium and thorium averages agree to within 20% of the surface-sample averages (Table 2 and Figure 3). Other insoluble elements include the high-field strength elements (HFSEs – titanium, zirconium, hafnium, niobium, tantalum, molybdenum, tungsten), beryllium, aluminum, gallium, germanium, indium, tin, lead, and a number of transition metals (chromium, cobalt, nickel, copper, and zinc). Taylor and McLennan (1985) noted that some of these insoluble elements (e.g., HFSEs) may be fractionated during sedimentary processing if they reside primarily in heavy minerals. More recent evaluations have suggested that this effect is probably not significant for niobium and tantalum, and fractionations of zirconium and hafnium due to heavy mineral sorting are only really apparent in loess (Barth et al., 2000; McLennan, 2001b; Plank and Langmuir, 1998). Plank and Langmuir (1998) noted that the niobium, tantalum, and titanium concentrations derived for the upper crust using

11

marine sediments are considerably different from those of Taylor and McLennan’s upper continental crust composition. As oceanic processes are unlikely to fractionate these elements, Plank and Langmuir (1998) suggested that marine sediments provide a reliable estimate of the average composition of the upper continental crust. Using correlations between Al2O3 and niobium, they derived a niobium concentration for the upper crust of 13.7 ppm, and tantalum of 0.96 ppm (assuming Nb/ Ta ¼ 14); these values are about a factor of 2 lower than Taylor and McLennan’s (1985) upper crustal estimates. Taylor and McLennan (1985) adopted their niobium value from Shaw et al. (1976), and their tantalum value was derived by assuming a Nb/Ta ratio of
12 for the upper crust. The Plank and Langmuir niobium and tantalum values are similar to those derived from the surface-sampling studies of Condie (1993), Gao et al. (1998a), and Borodin (1998) and from more recent evaluations of these elements in shales, loess and other terrigenous sedimentary rocks (Barth et al., 2000; Gallet et al., 1998; McLennan, 2001b). All of these estimates range between 10 ppm and 14 ppm niobium and 0.74 ppm to 1.0 ppm tantalum, an overall variation of
30%. Thus, niobium and tantalum concentrations now appear to be nearly as well constrained as the REE in the upper continental crust. Plank and Langmuir (1998) also suggested, from their analyses of marine sediments, increasing the upper crustal TiO2 values by
40% (from 0.5 wt% to 0.76 wt%). Thus the TiO2 content of the upper-continental crust probably lies between 0.55 wt% and 0.76 wt%, a difference of
30%. Of the remaining insoluble elements, recent evaluation of zirconium and hafnium concentrations derived from terrigenous sediment (McLennan, 2001b) show no significant differences with Taylor and McLennan’s estimates, whose upper crustal zirconium value derives from the Handbook of Geochemistry (Wedepohl, 1969–1978), with hafnium determined from an assumed Zr/Hf ratio of 33. These values lie within
20% of the surface-exposure averages (Table 2, Figure 3). For the insoluble transition metals chromium, cobalt, and nickel, McLennan’s (2001b) recent evaluation suggests approximate factor of 2 increases in average upper crustal values over those of Taylor and McLennan (1985). Taylor and McLennan’s (1985) values were taken from a variety of sources (see Table 1 of Taylor and McLennan, 1981) and are similar to the Canadian Shield averages, which appear to represent a more felsic upper-crust composition, as discussed above. Even after eliminating these lower values, 30–40% variation exists for chromium, cobalt, and nickel between different estimates (Table 2 and Figure 3), and the upper crustal concentrations of these elements remains poorly constrained relative to REE. McLennan (2001b) evaluated the upper crustal lead concentration from sediment averages and suggested a slight (
15%) downward revision (17 ppm) from the value of Taylor and McLennan (1985), whose value derives from a study by Heinrichs et al. (1980). McLennan’s value is identical to that of surface averages (Table 2) and collectively these should be considered as a robust estimate for the lead content of the upper crust. For the remaining insoluble elements – beryllium, copper, zinc, gallium, germanium, indium, and tin – no newer data are available for terrigenous sediment averages. Estimates for some elements (e.g., zinc, gallium, germanium, and indium) vary by only
2030% between different

12

Composition of the Continental Crust

studies, but others (beryllium, copper, and tin) vary by a factor of 2 or more (Table 2 and Figure 3). It may also be possible to derive average upper crustal abundances of elements that have intermediate solubilities (e.g., vanadium, arsenic, silver, cadmium, antimony, caesium, barium, tungsten, and bismuth) using their concentrations in fine-grained sedimentary rocks, if they show significant correlations with lanthanum. Using this method McLennan (2001b) derived estimates of the upper crustal composition for barium (550 ppm) and vanadium (107 ppm). McLennan’s barium value does not differ from that of Taylor and McLennan (1985), which derives from the Handbook of Geochemistry (Wedepohl, 19691978). This value is
10% to a factor of 2 lower than the shield estimates; 630700 ppm seems to be the most common estimate for barium from surface exposures. McLennan’s vanadium estimate is
50% higher than that of Taylor and McLennan (1985), which was derived from a 50 : 50 mixture of basalt : tonalite (Taylor and McLennan, 1981) and is similar to the Canadian Shield averages (Table 2). The revised vanadium value is similar to the surface-exposure averages from eastern China (Gao et al., 1998a) and Condie’s (1993) global average. Several studies have used data for sedimentary rocks to derive the concentration of caesium in the upper crust. McDonough et al. (1992) found that a variety of sediments and sedimentary rocks (including loess) have an Rb/Cs ratio of 19 (11, 1s), which is lower than the value of 30 in Taylor and McLennan’s (1985) upper crust. Using this ratio and assuming a rubidium content of 110 ppm (from Shaw et al., 1986; Shaw et al., 1976; Taylor and McLennan, 1981), led them to an upper crustal caesium concentration of
6 ppm. Data for marine sediments compiled by Plank and Langmuir (1998) also support a lower Rb/Cs ratio of the upper crust. Using the observed Rb/Cs ratio of 15 and a rubidium concentration of 112 ppm, they derived an upper crustal caesium concentration of 7.8 ppm. Although caesium data show only a poor correlation with lanthanum, the apparent La/Cs ratio of sediments led McLennan (2001b) to a revised caesium estimate of 4.6 ppm, which yields an Rb/Cs ratio of 24. Very few data exist for caesium from shield composites. Gao et al. determined a value of 3.6 ppm caesium, which is very similar to the estimate of Taylor and McLennan (1985). However, the Gao et al. rubidium estimate (83 ppm) is lower than Taylor and McLennan’s (112 ppm), leading to an Rb/Cs ratio of 23 in the upper crust of eastern China. Caesium concentrations in all estimates vary by up to 70% and there thus appears to be substantial uncertainty in the upper crust’s caesium concentration. Further evidence for the caesium content of the upper continental crust is derived from loess (see next section). The upper crustal abundances of arsenic, antimony, and tungsten were determined by Sims et al. (1990), based on measurements of these elements in loess and shales. They find As/Ce to be rather constant at 0.08, leading to an arsenic content of 5.1  1 ppm. In a similar fashion they estimate the upper crustal antimony content to be 0.45  0.08 ppm and tungsten to be 3.3  1.1 ppm. The antimony and arsenic values are factors of 2 and 3 higher, respectively, than the values given by Taylor and McLennan (1985), and the tungsten contents are a factor of 2 lower than Taylor and McLennan’s (1985), which were adopted from the Handbook of Geochemistry (Wedepohl,

1969–1978). For all three elements, the Sims et al. estimates lie within uncertainty of the values given by Gao et al. (1998a) for the upper crust of eastern China, and these new estimates can thus be considered as representative of the upper crust to within
30% uncertainty. For the remaining moderately soluble elements silver, cadmium, and bismuth, there are no data for sedimentary composites. Taylor and McLennan (1985) adopted values from Heinrichs et al. (1980) for cadmium and bismuth and from the Handbook of Geochemistry (Wedepohl, 1969–1978) for silver. The only other data come from the study of Gao et al. (1998a). So essentially there are only two studies that address the concentrations of these elements in the upper crust: Gao et al. (1998a) and Wedepohl (1995) (which incorporates data from the Handbook of Geochemistry and Heinrichs et al. (1980)). For silver and cadmium, the two estimates converge: silver is identical and cadmium varies by 25% between Gao et al. and Wedepohl et al. estimates. In contrast, bismuth shows a factor of 2 variation, with the Gao et al. estimates being higher.

4.1.2.2.2 Glacial deposits and loess The concept of analyzing glacial deposits in order to determine average upper crustal composition originated with Goldschmidt (1933, 1958). The main attraction of this approach is that glaciers mechanically erode the rock types that they traverse, giving rise to finely comminuted sediments that represent averages of the bedrock lithologies. Because the timescale between erosion and sedimentation is short, glacial sediments experience little chemical weathering associated with their transport and deposition. In support of this methodology for determining upper crust composition, Goldschmidt noted that the major-element composition of composite glacial loams from Norway (analysed by Hougen et al., 1925, as cited in Goldschmidt, 1933, 1958), which sample
2  105 km2 of Norwegian upper crust, compares favorably with the average igneous-rock composition determined by Clarke and Washington (1924) (Table 1). It would take another fifty years before geochemists returned to this method of determining upper crustal composition. More recent studies using glacial deposits to derive average upper-crust composition have focused on the chemical composition of loess – fine-grained eolian sediment derived from glacial outwash plains (Taylor et al., 1983; Gallet et al., 1998; Peucker-Ehrenbrink and Jahn, 2001; Hattori et al., 2003). This can be accomplished in two ways: either using the average composition of loess as representative of the upper continental crust or, if an element correlates with an insoluble element such as lanthanum whose upper concentration is well established, using the average X/La ratio of loess (where ‘X’ is the element of interest), and assuming an upper crustal lanthanum value to determine the concentration of ‘X’ (e.g., McLennan, 2001b). In this and subsequent discussion of loess, we derive upper crustal concentrations for particular elements using this method and assuming an upper crustal lanthanum value of 31 ppm, and compare these to previous estimates for these elements. The quoted uncertainty reflects 1s on that ratio. Loess is rich in SiO2 (most carbonate-free loess has 73 wt% to 80 wt% SiO2 (Taylor et al., 1983; Gallet et al., 1998), which probably reflects both the preferential eolian transport of quartz into loess and sedimentary recycling processes. This

Composition of the Continental Crust

enrichment causes other elemental concentrations to be diluted. In addition, some other elements may be similarly fractionated during eolian processing. For example, loess shows anomalously high concentrations of zirconium and hafnium (Taylor et al., 1983; Barth et al., 2000), which, like the SiO2 excess, have been attributed to size sorting through eolian concentration of zircon (Taylor et al., 1983). Thus, loess Zr/La and Hf/La are enriched relative to the upper continental crust and cannot be used to derive upper crustal zirconium and hafnium concentrations. In addition, a recent study of rhenium and osmium in loess suggests that osmium contents are enhanced in loess compared to its source regions (Hattori et al., 2003). This is explained by Hattori et al. (2003) as being due to preferential sampling of the fine sediment fraction by the wind, which may be enriched in mafic minerals that are soft and hence more easily ground to finer, transportable particle sizes. Mafic-mineral enhancement could give rise to similar fractionations between lanthanum and elements that are found primarily in mafic minerals (e.g., nickel, vanadium, scandium, chromium, cobalt, manganese, etc.). In such cases neither averages nor La/X ratios can be used to determine a reliable estimate of upper crustal composition. However, it is not apparent that eolian processing has significantly fractionated incompatible elements from lanthanum (e.g., barium, strontium, potassium, rubidium, niobium, thorium, etc.) that are not hosted primarily in mafic minerals. Indeed, the close correspondence of the thorium content of the upper crust derived from loess La–Th correlations (10.5  1 ppm; Figure 6) to that deduced from shales (10.7 ppm, Taylor and McLennan, 1985) suggests that upper crustal concentrations of these elements derived from loess La–X correlations are not significantly affected by eolian processing. Taylor et al. (1983), and later Gallet et al. (1998), determined the trace-element composition of a variety of loess samples from around the world and found that their REE patterns are remarkably constant and similar to that of average shales (see previous section and Figure 5). Likewise, niobium, tantalum, and thorium show strong positive correlations with the REE (Figure 6; Barth et al., 2000; Gallet et al., 1998). Thus, it appears that loess provides a robust estimate of average upper crustal composition for insoluble, incompatible trace elements. Because loess is glacially derived, weathering effects are significantly reduced compared to shales (Taylor et al., 1983), raising the possibility that loess may provide robust upper crustal estimates for the more soluble trace elements. However, examination of the major-element compositions of loess shows that all bear the signature of chemical weathering (Gallet et al., 1998). Gallet et al. attributed this to derivation of loess particles from rocks that had previously experienced sedimentary differentiation. Likewise, Peucker-Ehernbrink and Jahn (2001) noted a positive correlation between 87Rb/86Sr and 87Sr/86Sr in loess, indicating that the weathering-induced fractionation is an ancient feature, and therefore inherited from the glacially eroded bedrocks. Even so, the degree of weathering in loess, as measured by the ‘chemical index of alteration’ (CIA ¼ molar Al2O3/(Al2O3 þ CaO þ Na2O þ K2O) Nesbit and Young (1984)), is small relative to that seen in shales (Gallet et al., 1998), and it is likely that loess would provide a better average upper crustal estimate for moderately

13

soluble trace elements (e.g., arsenic, silver, cadmium, antimony, caesium, barium, tungsten, and bismuth) than shales. Unfortunately, few measurements of these elements in loess are available (Barth et al., 2000; Gallet et al., 1998; Jahn et al., 2001; Taylor et al., 1983). Barium data show a scattered, positive correlation with lanthanum, yielding an upper crustal average of 510  139 ppm (Figure 6). This value of barium concentration is similar to the one adopted by Taylor and McLennan (1985) and is within the uncertainty of all the other estimates save those of the Canadian Shield, which are significantly higher (Table 2). Caesium also shows a positive, scattered correlation with lanthanum, yielding an uncertain upper crustal caesium content of 4.8  1.6 ppm, which is similar to that recently suggested by McLennan (2001b). However, caesium shows a better correlation with rubidium (Figure 7), defining an Rb/Cs ratio of
17 in loess. Thus, if the upper crustal rubidium concentration can be determined, better constraints on the caesium content can be derived. The highly soluble elements (lithium, potassium, rubidium, strontium, and uranium) show variable degrees of correlation with lanthanum in loess. Strontium shows no correlation with lanthanum, which is likely due to variable amounts of carbonate in the loess samples (Taylor et al., 1983). Teng et al. (2004) recently reported lithium contents and isotopic compositions of shales and loess. Lithium contents of loess show no correlation with lanthanum, but fall within a limited range of compositions (17–41 ppm), yielding an average of 29  10 ppm (n ¼ 14). A similar value is derived using the correlation observed between lithium and niobium in shales. Thus, Teng et al. (2004) estimated the upper crustal lithium content at 35  11 ppm, which is somewhat higher than most previous estimates (Shaw et al., 1976; Taylor and McLennan, 1985; Gao et al., 1998a). Potassium and rubidium show scattered, positive correlations with lanthanum (the Rb–La correlation is better, and the K–La is worse than the Ba–La correlation) (Figure 6). These correlations yield an upper crustal rubidium concentration of 84  17 ppm. This rubidium value is identical to those derived from surface sampling by Eade and Fahrig (1973) , Condie (1993) , and Gao et al. (1998a), but is lower than the widely used value of Shaw et al. (1976) at 110 ppm. The latter was adopted by both Taylor and McLennan (1985) and Wedepohl (1995) for their upper crustal estimates. The weak K–La correlation yields an upper crustal K2O value of 2.4  0.5 wt%. This is within error of the surface-exposure averages of Fahrig and Eade (1968), Condie (1993), and Gao et al. (1998a), but is lower than the Shaw et al. surface averages of the Canadian shield (Shaw et al., 1967), values for the Russian platform and the value adopted by Taylor and McLennan (1985) based on K/U and Th/U ratios. The loess-derived K/Rb ratio is 238, which is similar to the ‘well established’ upper crustal K/Rb ratio of 250 (Taylor and McLennan, 1985). Because both potassium and rubidium are highly soluble elements, and loess shows evidence for some weathering, the potassium and rubidium contents derived from loess are best viewed as minimum values for the upper crust. Uranium shows a reasonable correlation with lanthanum (Figure 6), which yields an upper crustal uranium content of 2.7  0.6 ppm. This value is within error of the averages

14

Composition of the Continental Crust

20 14

Th

Nb

12

15

10 10

8 Taylor and McLennan

6

2 10

15

20

25

30

35

40

5

Gao et al.

r 2 = 0.82

4

r 2 = 0.54

This study

10

15

20

25

30

35

40

800

45 120

Rb

Ba

700

100

600

80

500 60 400 40

300 r 2 = 0.41

r 2 = 0.29

200 100 10

15

20

25

30

35

40

10

15

20

25

30

35

40

4.0

20 45 3.5

K2O

U 3.0

3.0

2.5 2.0 2.0 1.0 r 2 = 0.15 0.0 10

15

20

25 30 La (ppm)

35

40

r 2 = 0.48 10

15

20

25 30 La (ppm)

35

40

1.5 1.0 45

Figure 6 Lanthanum versus moderately to highly soluble incompatible trace elements in loess. Although loess is derived in part from weathered source regions, the positive correlations suggest that weathering has not completely obliterated the original, upper crustal mixing trends. Lines represent linear fit to data. Various models for the average upper crustal composition are superimposed (Taylor and McLennan, 1995, as modified by McLennan, 2001b; Gao et al., 1998a and this study – Table 3). (sources Taylor et al., 1983; Gallet et al., 1998; Barth et al., 2000; Jahn et al., 2001; Peucker-Ehernbrink and Jahn, 2001).

Cs (ppm)

8.0

Loess

Rb/Cs = 10

6.0

20 4.0

2.0

r 2 = 0.36 0.0

40

60

80

100

120

Rb (ppm) Figure 7 Rubidium versus caesium concentrations in loess samples. Short line is linear regression of data, thin, labeled lines represent constant Rb/Cs ratios. Symbols for crustal models and data sources as in Figure 6.

derived from surface exposures, except for the value of Gao et al. (1998a) and Eade and Fahrig (1973), which are distinctly lower. The loess-derived K/U ratio of 7400 is lower than that assumed for the upper crust of 10 000 (Taylor and McLennan, 1985), and may reflect some potassium loss due to weathering, as discussed above. Peucker-Ehrenbrink and Jahn (2001) analyzed loess in order to determine the concentrations of the platinum-group element (PGE) in the upper continental crust. To do this they examined PGE-major-element trends and used previously determined major-element compositions of the upper continental crust to infer the PGE concentrations (Table 2). Of the elements analyzed (ruthenium, palladium, osmium, iridium, and platinum), they found positive correlations for ruthenium, palladium, osmium, and iridium with major and trace elements for which upper crustal values had previously been established, leading to suggested upper crustal abundances of

15

Composition of the Continental Crust

340 ppt, 520 ppt, 31 ppt, and 22 ppt, respectively. They found no correlation between platinum contents and other elements, and so they simply used the average loess platinum content (510 ppt) as representative of the upper continental crust. We have estimated uncertainty on these values (shown in Table 2) by using the 95% confidence limit on the correlations

Table 3

Recommended composition of the upper continental crust

Element

Units

Upper crust

SiO2 TiO2 Al2O3 FeOT MnO MgO CaO Na2O K2O P2O5 Li Be B N F S Cl Sc V Cr Co Ni Cu Zn Ga Ge As Se Br Rb Sr Y Zr Nb Mo Ru Pd

wt% ” ” ” ” ” ” ” ” ” mg g1 ” ” ” ” ” ” ” ” ” ” ” ” ” ” ” ” ” ” ” ” ” ” ” ” ng g1 ”

66.6 0.64 15.4 5.04 0.10 2.48 3.59 3.27 2.80 0.15 24 2.1 17 83 557 621 370 14.0 97 92 17.3 47 28 67 17.5 1.4 4.8 0.09 1.6 84 320 21 193 12 1.1 0.34 0.52

a

1 Sigma

%

1.18 0.08 0.75 0.53 0.01 0.35 0.20 0.48 0.23 0.02 5 0.9 8

2 13 5 10 13 14 6 15 8 15 21 41 50

56 322 382 0.9 11 17 0.6 11 4 6 0.7 0.1 0.5 0.05

10 53 103 6 11 19 3 24 14 9 4 9 10 54

17 46 2 28 1 0.3 0.02 0.02

20 14 11 14 12 28 6 3

published by Peucker-Ehrenbrink and Jahn (2001) and, for platinum, the standard deviation of the mean (Table 3). Recently, Hattori et al. (2003) suggested that preferential sampling of mafic minerals in loess may lead to enhancement of PGE and thus, loess-derived estimates may represent maximum concentrations for the upper crust. Based on samples of glacially

Sourcea

Element

Units

Upper crust

1 2 1 1 1 1 1 1 3 1 11 4 4 5 4 4 4 6 6 6 6 6 7 7 8 4 9 4 5 10 4 4 4 11 12 13 13

Ag Cd In Sn Sb I Cs Ba La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu Hf Ta W Re Os Ir Pt Au Hg Tl Pb Bi Th U

ng g1 mg g1 ” ” ” ” ” ” ” ” ” ” ” ” ” ” ” ” ” ” ” ” ” ” ” ng g1 ” ” ” ” mg g1 ” ” ” ” ”

53 0.09 0.056 2.1 0.4 1.4 4.9 628 31 63 7.1 27 4.7 1.0 4.0 0.7 3.9 0.83 2.3 0.30 1.96 0.31 5.3 0.9 1.9 0.198 0.031 0.022 0.5 1.5 0.05 0.9 17 0.16 10.5 2.7

1 Sigma

%

Sourcea

3 0.01 0.008 0.5 0.1 1.5 83 3 4

5 15 14 26 28 50 31 13 9 6

2 0.3 0.1 0.3 0.1

8 6 14 7 21

0.4 0.05 0.7 0.1 1

18 17 14 13 54

0.009 0.007 0.5 0.4 0.04 0.5 0.5 0.06 1.0 0.6

29 32 95 26 76 57 3 38 10 21

4 4 4 14 12 5 15 16 4 4 4 4 4 4 4 4 17 17 4 17 4 4 4 11 18 13 13 13 13 4 4 4 4 19 20 20

Sources: (1) Average of all surface exposure data from Table 1, excluding Shaw et al. (1967), which is replicated by Fahrig and Eade (1968). (2) As (1) above, but including sediment-derived data from Plank and Langmuir (1998) and McLennan (2001b). (3) As (1) above, but also including K2O value derived from loess (see text). (4) Average of all values in Table 2, excluding Wedepohl (1995) value or Taylor and McLennan (1985) value for Au, if it is derivitive from Shaw et al. (1976) and Taylor and McLennan (1985). (5) Wedepohl (1995). (6) Average of all surface composite data in Table 2, excluding Shaw et al. (1976), and including additional data from sediments (McLennan, 2001b). (7) Average of all surface composite data in Table 2, excluding Shaw et al. (1976), and including Taylor and McLennan (1985) values. (8) Average of all surface composite data in Table 2, excluding Shaw et al. (1976) due to their fractionated Ga/Al ratio. (9) Average of sedimentary data from Table 2 (Sims et al., 1990) and Gao et al. (1998a) surface averages. (10) Dervied from La/Rb correlation in loess (see text). Value is identical to surface exposure data except for the Shaw et al. (1976) values. Data from Handbook of Geochemistry are about a factor of 2 lower than the latter and are not included in the average. (11) Average of all surface exposure data in Table 2 (minus Shaw et al. (1976), values) plus data from sediments and loess (Plank and Langmuir, 1998; Barth et al., 2000; McLennan, 2001b; Teng et al., 2004). (12) Average of all data in Table 2, excluding Taylor and McLennan (1985), which derive from same source as Wedephol’s. (13) From Peucker-Ehrenbrink and Jahn (2001); see text for origin of error estimates. (14) Average of all data in Table 2, excluding Taylor and McLennan (1985), which is a factor of two higher than all other estimates. (15) Derived from Rb/Cs ¼ 17 and upper crustal Rb value (see text). (16) Average of all data in Table 2, excluding the Shaw et al. (1976) and including additional data from loess (see text). (17) Value interpolated from REE pattern. (18) Average of all values in Table 2, plus correlation from Newsom et al. (1996), assuming W/Th ¼ 0.2. (19) Average of all values in Table 2, excluding the Shaw et al. (1976) value, which is a factor of 5 lower than the others. (20) From loess correlations with La (see text). Both values are within error of the average of all surface exposure data and other sedimentary data (Taylor and McLennan, 1985; McLennan, 2001b).

16

Composition of the Continental Crust

derived desert sands and glacial moraines, Hattori et al. (2003) estimated an upper crustal osmium abundance of
10 ppt. Prior to these studies, few estimates were available for the PGE content of the upper continental crust. PeuckerEhrenbrink and Jahn’s loess-derived palladium value is similar to the value published by Taylor and McLennan (1985), which derives from the Handbook of Geochemistry (S. R. Taylor, personal communication), but is a factor of 3 smaller than that determined by Gao et al. (1998a) for the upper crust of eastern China. Peucker-Ehrenbrink and Jahn’s (2001) loess-derived osmium abundance is
65% lower than the estimate of Esser and Turekian (1993), which Peucker-Ehrenbrink and Jahn attribute to the hydrogenous uptake of osmium by the riverine sediments used in that study. Furthermore, the desert-sand and glacial moraine-derived osmium value of Hattori et al. (2003) is a factor of 3 lower than the estimate of PeuckerEhrenbrink and Jahn (2001). Peucker-Ehrenbrink and Jahn’s (2001) loess-derived iridium content is the same as that published for the Canadian Shield by Shaw et al. (1976). Thus, the upper crustal concentration of some PGE may be reasonably well constrained (e.g., palladium and iridium), while considerable uncertainty remains for others (e.g., platinum and osmium). Rhenium is a highly soluble element that is easily leached during weathering, so the rhenium abundances of loess cannot be used directly to infer its upper crustal abundance. Following Esser and Turekian (1993), Peucker-Ehrenbrink and Jahn (2001) used the average 187Os/188Os ratio, osmium concentration, and average neodymium-model age of the crust to calculate the rhenium content of the upper continental crust. Their value (198 ppt) is about half that reported in Taylor and McLennan (1985) and calculated by Esser and Turekian (1993), who used the higher osmium abundance in their calculation. Using a similar methodology and osmiumisotopic composition, and the lower osmium abundance determined for glacially derived desert sands, Hattori et al. (2003) determined an upper crustal 187Re/188Os ratio of 35, which (assuming an average neodymium model age of 2.2 Ga for the crust) corresponds to a rhenium content of 74 ppt, about a third of the concentration determined by PeuckerEhrenbrink and Jahn from loess data. Sun et al. (2003) used the rhenium contents of undegassed arc lavas to estimate the rhenium content of the bulk continental crust, assuming that the crust grows primarily by arc accretion. Their value of 2.0  0.1 ppb is over an order of magnitude higher than that estimated by Peucker-Ehrenbrink and Jahn (2001) and Hattori et al. (2003) and is
5 times higher than the Esser and Turekian (1993) and Taylor and McLennan (1985) values. Because rhenium is a moderately incompatible element, the rhenium concentration of the upper crust should be comparable to or higher than the bulk crust value (similar to ytterbium). However, this extreme rhenium concentration would require an order of magnitude higher osmium concentration in the crust or an extremely radiogenic crust composition, neither of which are consistent with any current estimates. Sun et al. (2003) suggest that rhenium may be lost from the continents by either rhenium degassing during arc volcanism or continental rhenium deposition into anoxoic sediments that are recycled into the mantle. Thus the value of 2 ppb rhenium is a maximum value for the upper continental

crust and our knowledge of the rhenium content of the upper crust remains uncertain.

4.1.2.3

An Average Upper-Crustal Composition

In Table 3 we present our best estimate for the chemical composition of the upper continental crust. The footnote provides detailed information on how the value for each element was derived. In general, major-element values represent averages of the different surface-exposure studies, and errors represent one standard deviation of the mean. Because two independent studies are available for the Canadian Shield, and because it appears the Canadian Shield has lower abundances of mafic lithologies and higher abundances of sodiumrich tonalitic–trondhjemitic granitic gneisses compared to other areas (see Section 4.1.2.1), we include only values from one of these studies – the Fahrig and Eade (1968) study, which encompasses a greater number of samples compared to that of Shaw et al. (1967). We also incorporate TiO2 values derived from recent sedimentary studies (McLennan, 2001b; Plank and Langmuir, 1998) and the K2O value from loess (Section 4.1.2.2) into the upper crustal averages (note that including the latter value in the average does not significantly change it). The standard deviation for most major-element averages is 10% or less. Only the ferromagnesian elements (iron, manganese, magnesium, and titanium), Na2O, and P2O5 vary by up to 15%. The trace-element abundances shown in Table 3 derive from different methods depending on their solubility. For most insoluble elements (see Section 4.1.2.2 and Figure 4 for definitions of solubility), we average the surface composites in addition to sediment or loess-derived estimates to derive the upper crustal composition. The uncertainty reported represents 1s from the mean of all estimates. For moderately and highly soluble elements, we use the data derived from loess, if the elements show correlations with lanthanum (r2 > 0.4), to infer their concentrations. In this case, the error represents the SD of the X/La ratio (where X is the element of interest). For elements that show no or only a poor correlation with lanthanum in loess and sediments (e.g., K2O, Li, Ba, and Sr), we use the average of surface composites and sedimentary data (if some correlations exist with lanthanum) to derive an average. In most cases, the loess or sediment-derived values are within error of the surface-composite averages and these are noted in the footnote. Caesium is a special case. The loess caesium data show a poor correlation with lanthanum, but good correlation with rubidium. We thus use the observed Rb/Cs ratio of 17 (which is similar to the previous determination of this ratio in sedimentary rocks (McDonough et al., 1992)) and the upper crustal rubidium concentration of 84 ppm to derive the caesium concentration of 4.9 ppm in the upper crust. The error on this estimate derives from the standard deviation of the Rb/Cs ratio. For some elements, only single estimates are available (e.g., bromine, nitrogen, iodine), and these are adopted as reported. The uncertainty of these estimates is likely to be very high, but there is no way to estimate uncertainty quantitatively with such few data. Remarkably, the SD on a large number of trace elements is below 20%, and the concentrations of a few (flourine, scandium, vanadium, cobalt, zinc, gallium, germanium, arsenic, yttrium, niobium, LREEs, tantalum, lead, and thorium) would appear to be known within
10% (Table 3).

Composition of the Continental Crust

However, in a number of these cases (e.g., flourine, cobalt, gallium, germanium, and arsenic), the small uncertainties undoubtedly reflect the fact that there have been few independent estimates made of the upper crust composition for these elements. It is likely that the true uncertainty for these elements is considerably greater than expressed in Table 3. The upper crustal composition in Table 3 has many similarities to the widely used estimate of Taylor and McLennan (1985, with recent revision by McLennan (2001b)), but also some notable differences (Figure 8). Most of the elements that vary by more than 20% from the estimate of Taylor and McLennan are elements for which new data are recently available and few data exist overall (i.e., beryllium, arsenic, selenium, molybdenum, tin, antimony, rhenium, osmium). However, a number of estimates exist for K2O, P2O5, and rubidium contents of the upper crust and our estimates are significantly lower (by 20–40%) than Taylor and McLennan’s upper crust. The difference in P2O5 may simply be due to rounding errors. Taylor and McLennan (1985) report P2O5 of 0.2 wt% versus 0.15 wt% in our and other estimates of the upper crust – Tables 2 and 3). Taylor and McLennan (1985) derived their upper crustal K2O indirectly from thorium abundances by assuming Th/U ¼ 3.8 and K/U ¼ 10 000. The resulting K2O value is the highest of any of the estimates (Table 2 and Figure 2). Likewise, Taylor and McLennan’s rubidium value was determined from their K2O content, assuming a K/ Rb ratio of the upper crust of 250. Their rubidium concentration matches the Canadian Shield value of Shaw et al. (1976), but is higher than all other surface-exposure studies, including Fahrig and Eade’s Canadian Shield estimate (Table 2 and Figure 3). In contrast, the remaining surface-exposure studies match the rubidium value we derived from the loess Rb–La correlation (Figure 6). We conclude that the upper crust may have lower potassium and rubidium contents than estimated by Taylor and McLennan (1985). This finding has implications for total crustal heat production (see Section 4.1.4 and Chapter 4.2).

Taylor & McLennan/Rudnick & Gao

2.0

4.1.3

17

The Deep Crust

The deep continental crust is far less accessible than the upper crust and consequently, estimates of its composition carry a greater uncertainty. Compared to the upper crust, the earliest estimates of the composition of the deep crust are relatively recent (i.e., 1950s and later) and derive from both seismological and geological studies. On the basis of observed isostatic equilibrium of the continents and a felsic upper crust composition, Poldervaart (1955) suggested a two-layer crust with granodioritic upper crust underlain by a basaltic lower crust. The topic of deep crustal composition doesn’t seem to have been considered again until
20 years later, when a series of works in the 1970 s and 1980 s made significant headway into the nature of the deep continental crust. On the basis of surface heat-flow, geochemical studies of high-grade metamorphic rocks and seismological data, Heier (1973) proposed that the deep crust is composed of granulite-facies rocks that are depleted in heatproducing elements. A similar conclusion was reached by Holland and Lambert (1972) based on their studies of the Lewisian complex of Scotland. Smithson (1978) used seismic reflections and velocities to derive both structure and composition of the deep crust. He divided the crust into three, heterogeneous regions: (1) an upper crust composed of supracrustal metamorphic rocks intruded by granites, (2) a migmatitic middle crust and (3) a lower crust composed of a heterogenous mixture of igneous and metamorphic rocks ranging in composition from granite to gabbro, with an average intermediate (dioritic) composition. This three-layer model of the crust survives today in most seismologically based studies. Weaver and Tarney (1980, 1981, 1984) derived a felsic and intermediate composition for the Archean middle and lower crust, respectively, based on studies of amphibolite to granulite-facies rocks exposed in the Lewisian complex, Scotland. R. W. Kay and S. M. Kay (1981) were one of the first to stress the importance of xenolith studies to unravelling deep-crustal composition. They highlighted the

Upper continental crust

Sn

Re

1.8 1.6

Os Be

Rb

P

1.4

Mo

K

1.2 1.0 0.8 Ti 0.6

Mn Se

0.4

Sb

As Ti Fe Mg Na P Be Sc Cr Ni Zn Ge Se Sr Zr Mo Ag In Sb Ba Ce Nd Eu Tb Ho Tm Lu Ta Re

0.2 Si

Al Mn Ca K

Li

B

Ir

Tl

Bi

U

V Co Cu Ga As Rb Y Nb Pd Cd Sn Cs La Pr Sm Gd Dy Er Yb Hf W Os Au Pb Th

Figure 8 Plot of upper crustal compositional estimate of Taylor and McLennan (1995) (updated with values from McLennan, 2001b), divided by recommended values from this study. Horizontal lines mark 20% variation. Most elements fall within the 20% bounds; elements falling beyond these bounds are labeled. Of the elements that differ by over 20%, potassium and rubidium are probably the most significant, since these elements are commonly analyzed to high precision in crustal rocks.

18

Composition of the Continental Crust

heterogeneous nature of the deep crust and suggested its composition should vary depending on tectonic setting, cautioning against the use of singular cross sections or deep-crustal exposures to derive global models. Taylor and McLennan (1985) considered the lower crust to be the portion of the crust from 10 km depth to the Moho. Their ‘lower crust’ thus includes both middle and lower crust, as used here (see Section 4.1.3.1). Taylor and McLennan’s (1985) lower-crust composition was derived by subtracting the upper crust from their total-crust composition (see Section 4.1.4). The Taylor and McLennan (1985) lower crust is thus not based on observed lower crustal rock compositions, but rather on models of upper- and total-crust compositions and assumptions about the origin of surface heat flow. More recent attempts to define deep crust composition have relied upon linking geophysical data (principally seismic velocities) to deep crustal lithologies and their associated compositions to derive the bulk composition of the deep crust as a function of tectonic setting (Christensen and Mooney, 1995; Rudnick and Fountain, 1995; Wedepohl, 1995; Gao et al., 1998a,b). Despite the attendant large uncertainties in deriving composition from velocity (Rudnick and Fountain, 1995; Brittan and Warner, 1996, 1997; Behn and Kelemen, 2003) and the lack of thorough geochemical sampling of the deep crust in many regions, these efforts nevertheless provide the best direct estimates of present-day deep crustal composition. In this section we examine the composition of the deep crust by first defining its structure and lithology and the methods employed to determine deep crust composition. We then examine observations on middle and lower crustal samples, average seismic velocities and the resulting models of deep crust composition.

4.1.3.1

Definitions

Following recent compilations of the seismic-velocity structure of the continental crust, we divide the deep crust into middle and lower crust (Holbrook et al., 1992; Christensen and Mooney, 1995; Rudnick and Fountain, 1995). Holbrook et al. (1992) defined the middle crust as: (1) the middlethird, where the velocity structure suggests a natural division of the crust into thirds; (2) the region beneath the upper crust and above a Conrad discontinuity, if there is a layer beneath the Conrad; and (3) the region immediately beneath the Conrad if there are two distinct velocity layers beneath a Conrad discontinuity. The lower crust is thus the layer beneath the middle crust and above the Moho. For a
40 km thick average global continental crust (Christensen and Mooney, 1995; Rudnick and Fountain, 1995), the middle crust is 11 km thick and ranges in depth from 12 km, at the top, to 23 km at the bottom (Gao et al. (1998b) based on the compilations of data for crustal structure in various tectonic settings by Rudnick and Fountain (1995)). The average lower crust thus begins at 23 km depth and is 17 km thick. However, the depth and thickness of both middle and lower crust vary from setting to setting. In fore-arcs, active rifts, and rifted margins, the crust is generally thinner: middle crust extends from 8 km to 17 km depth and lower crust from 17 km to 27 km depth. In Mesozoic–Cenozoic orogenic belts the crust is thicker and middle crust extends from 16 km to

27 km depth and the lower crust from 27 km to 51 km depth (Rudnick and Fountain, 1995).

4.1.3.2

Metamorphism and Lithologies

Studies of exposed crustal cross-sections and xenoliths indicate that the middle crust is dominated by rocks metamorphosed at amphibolite facies to lower granulite facies, while the lower crust consists mainly of granulite facies rocks (Fountain et al., 1990a; Fountain and Salisbury, 1981; Mengel et al., 1991; Weber et al., 2002). However, exceptions to these generalities do occur. For thin crust in rifted areas, greenschistfacies and amphibolite-facies rocks may predominate in the middle and lower crust, respectively. In overthickened Mesozic and Cenozoic orogenic belts (e.g., Alps, Andes, Tibet, and Himalyas), and paleo-orogenic belts that now have normal crustal thicknesses (e.g., Appalachains, Adirondacks, Variscan belt), granulite-facies and eclogite-facies rocks may be important constituents of the middle and lower crust (Leech, 2001; LePichon et al., 1997; Lombardo and Rolfo, 2000). In contrast, amphibolite-facies lithologies may be present in the deep crust of continental arcs (Aoki, 1971; Miller and Christensen, 1994; Weber et al., 2002), where hydrous fluids are fluxed from the subducting slab and the water contents of underplating magmas are high. Lithologically, both middle and lower crust are highly heterogeneous, as seen in surface exposures of high-grade metamorphic rocks, crustal cross-sections, and deep-crustal xenolith suites. However, there is a general tendency for the middle crust to have a higher proportion of evolved rock compositions (as observed in cross-sections and granulite-facies terranes) while the lower crust has a higher proportion of mafic rock types (as observed in xenolith suites (Bohlen and Mezger, 1989)). Metasedimentary lithologies are often present, albeit in small proportions. The exact proportions of felsic to mafic lithologies in the deep crust varies from place to place and can only be established through the study of crustal cross-sections or inferred from seismic velocity profiles of the crust (Christensen and Mooney, 1995; Rudnick and Fountain, 1995; Wedepohl, 1995; Gao et al., 1998b).

4.1.3.3

Methodology

There are three approaches to derive the composition of the deep crust (see Rudnick and Fountain (1995) for a review). 1. By studying samples derived from the deep crust. These occur as surface outcrops of high-grade metamorphic terranes (e.g., Bohlen and Mezger, 1989; Harley, 1989), tectonically uplifted crustal cross-sections (e.g., Fountain and Salisbury, 1981; Percival et al., 1992), and as deep-crustal xenoliths carried in volcanic pipes (Rudnick, 1992; Downes, 1993). 2. By correlating seismic velocities with rock lithologies (Christensen and Mooney, 1995; Rudnick and Fountain, 1995; Wedepohl, 1995, Gao et al., 1998a,b). 3. From surface heat-flow measurements (see Chapter 4.2). As pointed out by Jaupart and Mareschal (see Chapter 4.2), surface heat flow is the only geophysical parameter that is a direct function of crustal composition. In general, however, heat flow provides only very broad constraints on deep-crust composition due to the ambiguity involved in distinguishing the amount of surface heat flow arising from crustal

Composition of the Continental Crust radioactivity versus the Moho heat flux (see Chapter 4.2); Rudnick et al., 1998). Most models of the deep-crust composition fall within these broad constraints. The exception is the global model of Wedepohl, 1995, which produces more heat than the average surface heat flow in the continents, thereby allowing no mantle heat flux into the base of the crust (Rudnick et al., 1998). In addition, the regional model of Gao et al. (1998a) for eastern China produces too much heat to be globally representative of the continental crust composition (see discussion in Rudnick et al. (1998) and Jaupart and Mareschal (Chapter 4.2)). However, the Gao et al. composition may be representative of the continental crust of eastern China, where the crust is relatively thin (30–35 km) and the heat flow is high (>60 mW m2). In the remaining discussion of deep-crust composition, we rely most heavily on methods (1)–(2), above, but return to the question of heat flow when considering the bulk crust composition in Section 4.1.4. In addition to mineralogy, which is in turn a function of bulk composition and metamorphic grade, factors affecting the seismic velocities of the continental crust include temperature, pressure, and the presence or absence of volatiles, fractures, and mineralogical anisotropy. It is generally assumed that cracks and fractures are closed under the ambient confining pressures of the middle to lower crust (0.4–1.2 GPa). In addition, although evidence for volatile transport is present in many rocks derived from the deep crust (see Chapters 4.6 and 4.9), the low density of these fluids allows for their escape to the upper crust shortly after their formation. Hence, most studies assume the deep crust does not, in general, contain an ambient, free volatile phase (Yardley, 1986). Some minerals are particularly anisotropic with respect to seismic-wave speeds (e.g., olivine, sillimanite, mica (Christensen, 1982)), which can lead to pronounced seismic anisotropy in rocks if these minerals are crystallographically aligned through deformational processes (Meltzer and Christensen, 2001). This, in turn, could lead to over- or underestimation of representative seismic velocities of the deep crust if deformed rocks with such anisotropic minerals occur there. Olivine is not commonly stable in the deep crust, but other strongly anisotropic minerals are (e.g., mica, which is predominantly stable in the middle crust, and sillimanite, which is found in metapelitic rocks in the middle-to-lower crust). Some of the largest seismic anisotropies have been recorded in mica schists and gneisses, which can have average anisotropies over 10% (Christensen and Mooney, 1995; Meltzer and Christensen, 2001). Amphibole is also anisotropic and the average anisotropies for amphibolite are also
10% (Christensen and Mooney, 1995; Kern et al., 1996). In general, anisotropy is expected to be highest in metapelitic rocks and amphibolites, which contain the highest proportions of anisotropic minerals. These lithologies appear to be subordinate in middle-crustal sections and outcrops (described in the next section) compared to felsic gneisses, which typically have low anisotropies (<5%). In contrast, studies of xenoliths show metapelite to be a common lithology in the lower crust, albeit proportionally minor, and amphibolite may be important in some regions (Section 4.1.3.5.1). Thus, seismic anisotropy could be especially important in regions having large amounts of metasedimentary rocks (e.g., accretionary wedges) and

19

amphibolite (arc crust?) in the deep crust, but is less likely to be important in crust dominated by felsic metaigneous rocks or mafic granulites. Changes in P-wave velocity of a rock as a function of temperature and pressure are generally assumed to be on the order of 4  104 km s–1  C1 and 2  104 km s1 MPa1 (see Rudnick and Fountain (1995) and references therein). Because most laboratory measurements of ultrasonic velocities are carried out at confining pressures of 0.6–1.0 GPa, no pressure correction needs to be made in order to compare field and laboratory-based velocity measurements. However, temperature influence on seismic-wave speeds can be significant, especially when comparing laboratory data collected at room temperature to field-based measurements in areas of high heat flow (e.g., rifts, arcs, extentional settings). The decrease in compressional wave velocities in the deep crust under these high geotherms can be as much as 0.3 km s1 (see Rudnick and Fountain, 1995, figure 1). For these reasons, Rudnick and Fountain (1995) used regional surface heat flow and assumed a conductive geothermal gradient, to correct the field-based velocities to room-temperature conditions. In this way, direct comparisons can be made between velocity profiles and ultrasonic velocities of lower-crustal rock types measured in the laboratory. Another benefit of this correction is that deep-crustal velocities from areas with grossly different geotherms can be considered directly in light of possible lithologic variations. In subsequent sections we quote deep-crustal velocities corrected to room-temperature conditions as ‘temperature-corrected velocities.’

4.1.3.4

The Middle Crust

4.1.3.4.1 Samples The best evidence for the compositional makeup of the middle crust comes from studies of high-grade metamorphic terranes and crustal cross-sections. There are far fewer studies of amphibolite-facies xenoliths derived from mid-crustal depths (Grapes, 1986; Leeman et al., 1985; Mattie et al., 1997; Mengel et al., 1991; Weber et al., 2002) compared to their granulitefacies counterparts. This may be due to the fact that it can be difficult to distinguish such xenoliths from the exposed or near-surface amphibolite-facies country rocks through which the xenolith-bearing volcanic rocks erupted. For this reason, xenolith studies have not been employed to any large extent in understanding the composition of the middle crust, and most information about the middle crust comes from studies of high-grade terranes, crustal cross-sections, and seismic profiles. Interpreting the origin of granulite-facies terranes and hence their significance towards determining deep-crustal composition depends on unraveling their pressure–temperature–time history (see Chapters 4.7 and 4.8). Those showing evidence for a ‘clockwise’ P–T path (i.e., heating during decompression) are often interpreted as having been only transiently in the lower crust; they represent upper crustal assemblages that passed through high P–T conditions on their way back to the surface during continent-scale collisional orogeny. In contrast, granulite terrains showing evidence for isobaric cooling can have extended lower-crustal histories, and thus may shed light on deep-crustal composition (see discussion in Rudnick and Fountain (1995)). Bohlen and Mezger (1989) pointed out

20

Composition of the Continental Crust

that isobarically cooled granulite-facies terranes show evidence of equilibration at relatively low pressures (i.e., 0.6– 0.8 GPa), corresponding to mid-crustal depths (25 km). Although a number of high-pressure and even ultra-high-pressure metamorphic belts (Chapter 4.9) have been recognized since their study, it remains true that the majority of isobarically cooled granulite-facies terranes show only moderate equilibration depths and, therefore, may provide evidence regarding the composition of the middle crust. Although lithologically diverse, the average composition of rocks analyzed from granulite terrains is evolved (Rudnick and Presper, 1990), with median compositions corresponding to granodiorite/dacite (64–66 wt% SiO2, 4.1–5.2 wt% Na2O þ K2O, based on classification of Le Bas and Streckeisen (1991)). Rudnick and Fountain (1995) suggested that isobarically cooled granulite terrains have a higher proportion of mafic lithologies than granulites having clockwise P–T paths. However, the median composition of rocks analyzed from isobarically cooled terranes is indistinguishable (62 wt% SiO2, 4.6 wt% Na2O þ K2O) from the median composition of the entire granulite-terrane population given in Rudnick and Presper (1990). Collectively, these data point to a chemically evolved mid-crustal composition. Observations from crustal cross-sections also point to an evolved mid-crust composition (Table 4). Most of these crosssections have been exposed by compressional uplift due to thrust faulting (e.g., Kapuskasing, Ivrea, Kohistan, and Musgrave). Other proposed origins for the uplift include wide, oblique transitions (Pikwitonei), impactogenesis (Vredefort), and transpression (Sierra Nevada) (Percival et al., 1992). In nearly all these sections, sampling depth ranges from upper to middle crust; only a few (e.g., Vredefort, Ivrea, Kohistan) appear to penetrate into the lower crust. In the following paragraphs we review the insights into middle (and lower) crust lithologies gained from the studies of these crustal cross-sections. The Vredefort dome represents a unique, upturned section through
36 km of crust of the Kaapvaal craton, possibly exposing a paleo-Moho at its base (Hart et al., 1981, 1990; Tredoux et al., 1999; Moser et al., 2001). The origin of this structure is debated, but one likely scenario is that it was produced by crustal rebound following meteorite impact. The shallowest section of basement (corresponding to original depths of 10–18 km depth) is composed of amphibolite-facies rocks consisting of granitic gneiss (the outer granite gneiss). The underlying granulite-facies rocks (original depths of 1836 km) are composed of charnockites and leucogranofels with
10% mafic and ultramafic granulites (the Inlandsee Leucogranofels terrain). The mid-crust, as defined here, is thus composed of amphibolite-facies felsic gneisses in fault contact with underlying charnockites and mixed felsic granulites and mafic/ultramafic granulites (Hart et al., 1990). The lower crust, which is only partially exposed, consists of mixed felsic and mafic/ultramafic granulites, with the proportion of mafic rocks increasing with depth. The mantle beneath the proposed paleo-Moho, as revealed by borehole drilling, is dominated by 3.3–3.5 Ga serpentinized amphibole-bearing harzburgite (Tredoux et al., 1999). The Kapuskasing Structural Zone represents an exposed middle-to-lower crustal section through a greenstone belt of

the Archean Canadian Shield, where the middle crust is represented by the amphibolite-facies Wawa gneiss dome and lower granulite-facies litihologies along the Kapuskasing uplift. Altogether,
25 km of crust are exposed out of a total crustal thickness of 43 km (Fountain et al., 1990b; Percival and Card, 1983). The Wawa gneiss dome is dominated by tonalite– granodiorite gneisses and their igneous equivalents (87%), but also contains small amounts of paragneiss (5%) and mafic gneiss and intrusives (8%) (Burke and Fountain, 1990; Fountain et al., 1990b; Shaw et al., 1994). The slightly deeperlevel Kapuskasing Structural Zone has a greater proportion of paragneisses and mafic lithologies. It contains 35% mafic or anorthositic gneisses, 25% dioritic gneisses, 20% paragneiss, and only 20% tonalite gneisses. Like the high-grade rocks of the Kapusksasing Structural Zone, those in the Pikwitonei crustal cross-section represent high-grade equivalents of granite–gneiss–greenstone successions (Fountain and Salisbury, 1981; Percival et al., 1992). Approximately 25 km of upper-to-middle crust is exposed in this section out of a total-crustal thickness of 37 km (Fountain et al., 1990b). Both amphibolite- and granulite-facies rocks are dominated by tonalitic gneiss with minor mafic gneiss, and metasedimentary rocks. The Wutai–Jining Zone is suggested to be an exposed crosssection through the Archean North China craton (Kern et al., 1996). Rocks from this exposure equilibrated at depths of up to
30 km, thus sampling middle and uppermost lower crust, but leaving the lowermost 10 km of crust unexposed (Kern et al., 1996). Like the previously described cross-sections, felsic gneisses dominate the middle crust; tonalitic–trondhjemitic–granodioritc, and granitic gneisses comprise 89% of the dominant amphibolite to granulite-facies Henshan-Fuping terrains, the remaining lithologies are amphibolite-mafic granulite (8%) and metapelite (3%). Tonalitic–trondhjemitic–granodioritc and granitic gneiss (54%) are less significant but still dominant in the lower-crustal Jining terrain. The Musgrave Range (Fountain and Salisbury, 1981 ; Percival et al., 1992) and the Bamble Sector of southern Norway (Pinet and Jaupart, 1987; Alirezaei and Cameron, 2002) represent two crustal sections through Proterozoic crust of central Australia and the Baltic Shield, respectively. In both sections, the middle crust is dominated by quartzofeldspathic gneiss. The lower crust consists of silicic to intermediate gneiss, felsic granulite, and mafic granulite with layered mafic and ultramafic intrusions being important lithological components in the Musgrave Range and metasediments being important in the lower crust of southern Norway. The Ivrea–Verbano Zone in the southern Alps of Italy was the first to be proposed as an exposed deep-crustal section by Berckhemer (1969) and has subsequently been the focus of extensive geological, geochemical, and geophysical studies (e.g., Mehnert, 1975; Fountain, 1976; Dostal and Capedri, 1979; Voshage et al., 1990; Quick et al., 1995). The Paleozoic rocks of the Ivrea zone are unusual when compared with Precambrian granulite outcrops because they contain a large proportion of mafic lithologies and, as such, closely resemble granulite xenoliths in composition (Rudnick, 1990b). Amphibolite-facies rocks of the middle crust consist of felsic gneiss, amphibolite, metapelite (kinzigite), and marble, whereas the lower crustal section comprises mafic granulite and diorite, which formed by

Table 4

Chemical and petrological composition of crustal cross-sections Age

Setting, (uplift origin)b

Current crustal thickness (km)

Maximum depth (km)

Middle crust lithologies

Lower crust lithologies

1–3

2.6–3.6 Ga

Kaapvaal craton, (3)

36

Amphibolite-facies granitic gneiss

Kapuskasing Uplift

4–5

2.5–2.7 Ga

Superior craton, (1)

43

36 (w/ paleo Moho) 25

Pikwitonei granulite domain

6–9

2.5–3.1 Ga

Superior craton, (2)

37

25

Wutai-Jining zone

10

2.5–2.8 Ga

North China Craton, (2)

40

30

Amphibolite-lower granulite facies. Dominately tonalite gneiss, minor mafic gneisses, quartzites, anorthosites. Amphibolite-lower granuilte facies. 89% tonalitic-trondhjemiticgranodioritic-granitic gneiss 8% amphibolite and mafic granulite 3% metapelite

Granulite-facies charnockites, leucogranofels, mafic, and ultramafic granulites Lower granulite-facies 35% mafic/anorthositic 25% diorite 20% metasediments 20% felsic Granulite-facies. Predominantly silicic to intermediate gneiss, with minor paragneiss, mafic-ultramafic bodies and anorthosites Granulite-facies. 54% tonalitic-trondhjemiticgranodioritic-granitic gneiss 32% mafic granulite 6% metapelite 8% metasandstone

Proterozoic Musgrave ranges

6, 8, 11

1.1–2.0 Ga

Central Australia, (1)

40

Unknown

S. Norway

12–13

1.5–2.0 Ga

Baltic Shield, (2)

35

Unknown

14–17

Permian

Alps, (1)

35

30

8, 18–20

Cretaceous

Continental arc, (4)

27–43

30

8, 21

Late Jurassic–Eocene

Oceanic arc, (1)

Unknown

22–23

Jurassic

Oceanic arc, (1)

25–35

Archean Vredefort Dome

Phanerozoic IvreaVerbano zone Sierra Nevada, California Kohistan, Pakistan Talkeetna, Alaska

Amphibolite-facies 87% felsic 8% mafic-intermediate 5% metasediment

Quartzofeldspathic gneiss, amphibolite, metapelite, marble, calc-silicate gneiss Quartzofeldspathic gneiss, amphibolite, metasediments

Silicic to intermediate gneiss, mafic granulite, layered mafic-ultramafic intrusions Felsic granulite, mafic granulite, metasediments

Amphibolite-facies. felsic gneiss, amphibolite, metapelite (kinzigite), marble Mafic to felsic gneiss, amphibolite, diorite-tonalite

Granulite facies, mafic intrusives and ultramafic cumulates, resistic metapelite (stronalite), diorite Granofels, mafic granulite, graphite-bearing metasediments

45

Diroite, metadiorite, gabbronorite

13

Gabbro, tonalite, diorite

Amphibolite, metagabbro, gabbronorite, garnet gabbro, garnet homblendite, websterite Garnet gabbro, amphibole gabbro, dunite, wehrlite, pyroxenite

21

a References: 1. Hart et al., 1990, 2. Tredoux et al., 1999, 3. Moser et al., 2001, 4. Fountain et al., 1990a, 5. Shaw et al., 1994, 6. Fountain and Salisbury, 1981, 7. Fountain et al., 1987, 8. Percival et al., 1992, 9. Fountain and Salisbury 1995, 10. Kern et al., 1996, 11. Clitheroe et al., 2000, 12. Pinet and Jaupart, 1987, 13. Alirezaei and Cameron, 2002, 14. Mehnert, 1975, 15. Fountain, 1976, 16. Voshage et al., 1990, 17. Mayer et al., 2000, 18. Ross, 1985, 19. Saleeby, 1990, 20. Ducea, 2001, 21. Miller and Christensen, 1994, 22. Pearcy et al., 1990, 23. see Chapter 4.21 Magmatic arcs (Kelemen et al.). b The different mechanisms responsible for uplift of these crustal cross sections include (1) compressional uplifts along thrust faults, (2) wide, oblique transitions, which are also compressional in origin, but over wide transitions, with no one thrust fault obviously responsible for their uplift, (3) meteorite impact, and (4) transpressional uplifts, which are vertical uplifts along a transcurrent faults (Percival et al., 1992).

Composition of the Continental Crust

Referencea

22

Composition of the Continental Crust

intrusion and subsequent fractionation of basaltic melts that partially melted the surrounding metasediments (now resistic stronalite) (Mehnert, 1975; Dostal and Capedri, 1979; Fountain et al., 1976; Voshage et al., 1990). Detailed mapping by Quick et al. (1995) demonstrated that mantle peridotites in the southern Ivrea Zone are lenses that were tectonically interfingered with metasedimentary rocks prior to intrusion of the gabbroic complex and the present exposures reside an unknown distance above the pre-Alpine contiguous mantle. Thus reference to the section as a complete crust–mantle transition could be misleading. Altogether, the exposed rocks represent
30 km of crust with
5 km lowermost crust remaining unexposed (Fountain et al., 1990a). The similarity in isotope composition and age between the Ivrea zone cumulates and Hercynian granites in the upper crust led Voshage et al. (1990) to speculate that these granites were derived from lower-crustal magma chambers similar to those in the Ivrea Zone, suggesting that basaltic underplating may be important in the formation and modification of the lower continental crust (Rudnick, 1990a). Three sections through Mesozoic arcs show contrasting bulk compositions, depending on their settings (continental versus oceanic). In the southern Sierra Nevada, a tilted section exposes the deeper reaches of the Sierra Nevada batholith, which is part of a continental arc formed during the Mesozoic. This section is dominated by arc-related granitoids to depths of
30 km, which have a tonalitic bulk chemistry (Ducea and Saleeby, 1996; Ducea, 2001). At the deepest structural levels, the mafic Tehachapi Complex comprises mafic and felsic gneiss, amphibolite, diorite, tonalite, granulite, and rare metasediments (Percival et al., 1992; Ross, 1985). In contrast, two sections through accreted intraoceanic arcs have considerably more mafic middle-crust compositions. In the Jurassic Talkeetna section of southeastern Alaska, the middle crust comprises gabbro and tonalite (4.5 km), which is underlain by variably deformed garnet gabbro and gabbro with cumulate dunite, wehrlite, and pyroxenite (2.2 km) in the lower crust (Pearcy et al., 1990; see Chapter 4.21). The upper, middle, and lower crustal units are estimated to have an average SiO2 of 57%, 52%, and 44–45%, respectively. The Late Jurassic– Eocene Kohistan arc of Pakistan represents a 45 km thick reconstructed crustal column through a deformed, intruded intraoceanic arc sequence exposed in the Himalayan collision zone (Miller and Christensen, 1994). The depth interval from 10 km to 18 km is dominated by diorite and metadiorite. Rocks below this level, from
18 km to the Moho, are dominated by metamorphosed mafic to ultramafic rocks from a series of layered mafic intrusions. In summary, exposed amphibolite- to granulite-facies terranes and middle crustal cross-sections contain a wide variety of lithologies, including metasedimentary rocks, but they are dominated by igneous and metamorphic rocks of the diorite– tonalite–trondhjemite–granodiorite (DTTG), and granite suites. This is true not only for Precambrian shields but also for Phanerozoic crust and continental arcs, as documented in the crustal cross-sections described above. However, intraoceanic arcs may contain substantially greater proportions of mafic rocks in the middle and lower crust, as illustrated by the Kohistan and Talkeetna arc sections (Pearcy et al., 1990; Miller and Christensen, 1994; see Chapter 4.21).

4.1.3.4.2 Seismological evidence The samples described above provide evidence of the lithologies likely to be present in the middle crust. By definition, however, these samples no longer reside in the middle crust and additional information is required in order to determine the composition of the present-day middle crust. For this, we turn to seisomological data for continental crust from a variety of tectonic settings. Except for active rifts and some intraoceanic island arcs, which exhibit the highest middle-crust P-wave velocities (6.7  0.3 km s1 (Rudnick and Fountain, 1995) and 6.8 0.2km s1 (data from Holbrook et al., 1992) corrected to room temperature), other continental tectonic units have room-temperature middle-crustal P-wave velocities between 6.4 kms1 and 6.6 km s1 (Rudnick and Fountain, 1995). This range overlaps the average velocity of in situ middle crust, which was determined by Christensen and Mooney (1995) to be from 6.3 km s1 to 6.6  0.3 km s1, with an average of 6.5 0.2 km s1 over the depth range of 1525 km. When corrected for temperature (an increase of 0.1–0.2 km s1, depending on the regional geotherm), these average middle-crustal velocities are similar to the room-temperature velocities considered by Rudnick and Fountain (1995). Thus, the middle crust has average, room-temperature-corrected velocity between 6.4 km s1 and 6.7 km s1. Amphibolite-facies felsic gneisses have room temperature P-wave velocities of 6.4  0.1 km s1 (Rudnick and Fountain, 1995). This compares well with the room-temperature velocity of average biotite (tonalite) gneiss at 6.32  0.17 km s1 (20 km depth; Christensen and Mooney, 1995). Granitic gneiss has a slightly lower velocity (6.25  0.11 km s1; Christensen and Mooney (1995)), but is within uncertainty of the tonalite. A mixture of such gneisses with 0–30% amphibolite or mafic gneiss of the same metamorphic grade (Vp¼ 7.0 km s1; Rudnick and Fountain, 1995, Christensen and Mooney, 1995) yields P-wave velocities in the range observed for most middle crust. The above seismic data are thus consistent with the observations from granulite terranes and crustal cross-sections, and suggest that the middle crust is dominated by felsic gneisses.

4.1.3.4.3 Middle-crust composition Compared to other regions of the crust (upper, lower, and bulk), few estimates have been made of the composition of the middle crust (Table 5, and Figures 9 and 10). Moreover, these estimates provide data for a far more limited number of elements, and large differences exist between different estimates. The estimates of Weaver and Tarney (1984), Shaw et al. (1994) and Gao et al. (1998a) are based on surface sampling of amphibolite-facies rocks in the Lewisian Complex, the Canadian Shield, and Eastern China, respectively. Rudnick and Fountain (1995) modeled the middle crust as 45% intermediate amphibolite-facies gneisses, 45% mixed amphibolite and felsic amphibolite-facies gneisses, and 10% metapelite. This mixture is very similar to that of Christensen and Mooney (1995), who proposed a middle crust of 50% tonalitic gneiss, 35% amphibolite, and 15% granitic gneiss. Unfortunately, compositional data are not available for Christensen and Mooney’s samples and so the chemical composition of their middle crust cannot be calculated.

Composition of the Continental Crust

The estimates of Rudnick and Fountain (1995) and Gao et al. (1998a) show a broad similarity, although the latter is more evolved, having higher SiO2, K2O, barium, lithium, zirconium, and LREEs and LaN/YbN and lower total FeO, scandium, vanadium, chromium, and cobalt with a significant negative europium anomaly (Figures 9 and 10). These differences are expected, based on the slightly higher compressional velocity of Rudnick and Fountain’s global middle crust compared to that of Eastern China (6.6 km s1 versus 6.4 km s1; Gao et al., 1998b). The consistency is surprising considering that the two estimates are based on different sample bases and different approaches, one global and the other regional. The middle-crustal compositions of Weaver and Tarney (1984) and Shaw et al. (1994) deviate from the above estimates by being markedly higher in SiO2 and lower in TiO2, FeO, MgO, and CaO. Moreover, these middle-crust compositions are more felsic (based on the above elements) than all estimates of the upper-continental crust composition given in Table 1. Thus, it is unlikely that the Weaver and Tarney (1984) and Shaw et al. (1994) compositions are representative of the global average middle crust, as both heat flow and seismic observations require that the crust becomes more mafic with depth. It should be noted, however, that heat production for Shaw’s middle-crust composition is indistinguishable from those of Rudnick and Fountain (1995) and Gao et al. (1998a) at
1.0 mW m3, due largely to the very high K/Th and K/U of Shaw et al. estimate. The middle crust of Weaver and Tarney (1984) has significantly higher heat production, at 1.4 mW m3. Generally speaking, it would be best to derive the middlecrust composition from observed seismic-wave speeds and chemical analyses of amphibolite-facies rocks. However, few such data sets exist. Only two studies attempt to define the global average seismic-wave speeds for the middle crust (Christensen and Mooney, 1995; Rudnick and Fountain, 1995) and neither provides chemical data for amphibolite facies samples. Rudnick and Fountain used compiled chemical data for granulite-facies rocks and inferred the concentrations of fluid-mobile elements (e.g., rubidium, uranium) of their amphibolite facies counterparts, while Christensen and Mooney (1995) did not publish their chemical data for the amphibolitefacies rocks they studied. For this reason, we have chosen to estimate the middle-crust composition by averaging the estimates of Rudnick and Fountain (1995) and Gao et al. (1998a) (Table 5), where corresponding data are available. Although the latter study is regional in nature, its similarity to the global model of Rudnick and Fountain (1995) suggests that it is not anomalous from a global perspective (unlike the lower crust of Eastern China as described in Section 4.1.3.5) and it provides additional estimates for little-measured trace elements. This middle crust has an intermediate composition with lower SiO2 and K2O concentrations and higher FeO, MgO, and CaO concentrations than average upper crust (Table 1), consistent with the geophysical evidence (cited above) of a chemically stratified crust. Differences in trace-element concentrations between these two estimates are generally less than 30%, with the exceptions of P2O5, lithium, copper, barium, lanthanum, and uranium (Figure 10). The concentrations of these elements are considered to be less constrained. The middle crust is LREE enriched and exhibits the characteristic depletion of niobium

23

relative to lanthanum and enrichment of lead relative to cerium seen in all other parts of the crust (Figure 11). In summary, our knowledge of middle-crustal composition is limited by the small number of studies that have focused on the middle crust and the ambiguity in deriving chemical compositions from seismic velocities. Thus, the average composition given in Table 5 is poorly constrained for a large number of elements. Seismological and heat-flow data suggest an increase in seismic-wave speeds and a decrease in heat production with depth in the crust. Studies of crustal cross-sections show the middle crust to be dominated by felsic gneisses of tonalitic bulk composition. The average middle-crust composition given in Table 5 is consistent with these broad constraints and furthermore suggests that the middle crust contains significant concentrations of incompatible trace elements. However, the uncertainty on the middle-crust composition, particularly the trace elements, remains large.

4.1.3.5

The Lower Crust

4.1.3.5.1 Samples Like the middle crust, the lower crust also contains a wide variety of lithologies, as revealed by granulite xenoliths, exposed high-pressure granulite terranes and crustal crosssections. Metaigneous lithologies range from granite to gabbro, with a predominance of the latter in most lower crustal xenolith suites. Exceptions include xenolith suites from Argentina (Lucassen et al., 1999) and central Spain (Villaseca et al., 1999), where the xenoliths are dominated by intermediate to felsic granulites and the Massif Central (Leyreloup et al., 1977; Downes et al., 1990) and Hannuoba, China (Liu et al., 2001), where intermediate to felsic granulites comprise nearly half the population. Metapelites occur commonly in both terranes and xenoliths, but only rarely do other metasedimentary lithologies occur in xenolith suites; unique xenolith localities have been documented with meta-arenites (Upton et al., 1998) and quartzites (Hanchar et al., 1994), but so far marbles occur only in terranes. The reason for their absence in lower crustal xenolith suites is uncertain – they may be absent in the lower crust sampled by volcanoes, they may not survive transport in the hot magma, or they may simply have been overlooked by xenolith investigators. These issues related to the representativeness of xenolith sampling are the reason why robust estimates of lower-crustal composition must rely on a grand averaging technique, such as using seismic velocities to infer composition. Information on the lower crust derived from crustal crosssections has been given in Section 4.1.3.4.1 and only the main points are summarized here. All crustal cross-sections show an increase in mafic lithologies with depth and most of those in which possible crust-mantle boundaries are exposed reveal a lower crust that is dominated by mafic compositions. For example, in the Ivrea Zone, Italy, the lower crust is dominated by mafic granulite formed from basaltic underplating of country rock metapelite (Voshage et al., 1990). The same is true for the Kohistan sequence, Pakistan (Miller and Christensen, 1994), although here metapelites are lacking. Although the crust–mantle boundary is not exposed in the Wutai-Jining terrain, the granulite-facies crust exposed in this cross-section has a more mafic composition than the rocks of the middle-

24

Composition of the Continental Crust

Table 5 Compositional estimates of the middle continental crust. Major elements in weight percent. Trace element concentration units the same as in Table 2 1 Weaver and Tarney (1984) SiO2 TiO2 Al2O3 FeOTb MnO MgO CaO Na2O K2O P2O5 Mg# Li Be B N F S Cl Sc V Cr Co Ni Cu Zn Ga Ge As Se Br Rb Sr Y Zr Nb Mo Ru Pd Ag Cd In Sn Sb T I Cs Ba La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb

68.1 0.31 16.33 3.27 0.04 1.43 3.27 5.00 2.14 0.14 43.8

2 Shaw et al. (1994) 69.4 0.33 16.21 2.72 0.03 1.27 2.96 3.55 3.36 0.15 45.5 20.5

3 Rudnick and Fountain (1995) 62.4 0.72 15.96 6.59 0.10 3.50 5.25 3.30 2.07 0.10 48.6 7

3.2

32 20

74 580 9 193 6

713 36 69 30 4.4 1.09 0.41

0.14 0.76

5.4 46 43 30 18 8 50

22 118 83 25 33 20 70 17

92 465c 16 129 8.7 0.3

62 281 22 125 8

0.98 1376 22.9 42.1 18.3 2.8 0.78 2.11 0.28 1.54

0.63

2.4 402 17 45 5.8 24 4.4 1.5 4.0 0.58 3.8 0.82 2.3 2.3

4 Gao et al. (1998a)

5 This study a

64.6 0.67 14.08 5.45 0.11 3.67 5.24 3.48 2.52 0.19 54.5 16 2.29 17

63.5 0.69 15.0 6.02 0.10 3.59 5.25 3.39 2.30 0.15 51.5 12 2.29 17

524 20 182 15 95 69 18 34 32 69 18 1.13 3.1 0.064

524 20 182 19 107 76 22 33.5 26 69.5 17.5 1.13 3.1 0.064

67 283 17.0 173 11 0.60

65 282 20 149 10 0.60

0.76 48 0.061

0.76 48 0.061

1.30 0.28

1.30 0.28

1.96 661 30.8 60.3

2.2 532 24 53 5.8 25 4.6 1.4 4.0 0.7 3.8 0.82 2.3 0.32 2.2

26.2 4.74 1.20 0.76

2.17

1 Sigmaa

%

2 0.04 1 0.8 0.00 0.1 0.01 0.1 0.3 0.06

2 6 9 13 2 3 0 4 14 43

6

55

5 16 10 5 0.7 8 0.7 0.7

27 15 13 23 2 33 1 4

4 1 4 34 2

5 1 18 23 22

0.3 183 10 11

14 34 41 21

2 0.2 0.2

6 5 16

0.1

19

0.09

4

(Continued)

Composition of the Continental Crust

Table 5

(Continued) 1 Weaver and Tarney (1984)

Lu Hf Ta W Re Os Ir Pt Au Hg Tl Pb Bi Th U

25

2 Shaw et al. (1994)

0.1 3.8

3 Rudnick and Fountain (1995)

0.12 3.3 1.8

22 8.4 2.2

0.41 4.0 0.6

9.0

15.3

6.4 0.9

6.1 1.6

4 Gao et al. (1998a)

5 This study a

0.32 4.79 0.55 0.60

0.4 4.4 0.6 0.60

0.85 0.66 0.0079 0.27 15 0.17 6.84 1.02

0.85 0.66 0.0079 0.27 15.2 0.17 6.5 1.3

1 Sigmaa

%

0.06 0.6 0.04

17 13 6

0.2

1

0.5 0.4

8 31

a Averages and standard deviations of middle crustal composition by Rudnick and Fountain (1995) and Gao et al. (1998a), or from either of these two studies if data from the other one are unavailable. b Total Fe as FeO. c Recalculated from original data given by Shaw et al. (1994; Table 4), due to a typographical error in the published table. Mg# ¼ molar 100  Mg/(Mg þ Fetot). Major elements recast to 100% anhydrous.

Normalized to R and G

2.0

1.5

1.0

0.5

0.0

Weaver and Tarney Shaw et al. Gao et al. Rudnick and Fountain

Si

Al

Fe

Mg

Ca

Na

K

Figure 9 Comparison of the major-element composition of the middle continental crust as determined by sampling of surface exposures (Shaw et al., 1994; Weaver and Tarney, 1984) and inferred from middle-crustal seismic velocities combined with surface and xenolith samples (Rudnick and Fountain, 1995; Gao et al., 1998a). All values normalized to the new composition provided in Table 5 (‘R&G’), which is an average between the values of Gao et al. (1998a) and Rudnick and Fountain (1995). Gray shaded field represents 10% variation from this value.

crust section. Even in the Vredefort and Sierra Nevada crosssections, which are dominated by granitic rocks throughout most of the crustal sections (Ducea, 2001; Hart et al., 1990), the deepest reaches of exposed crust are characterized by more mafic lithologies (Ross, 1985; Hart et al., 1990; Table 4). There have been a number of studies of granulite-facies xenoliths since the reviews of Rudnick (1992) and Downes (1993) and a current tabulation of xenolith studies is provided in Table 6, which provides a summary of most lower crustal xenolith studies published through 2002. Perhaps most significant are the studies of lower-crustal xenoliths from

Archean cratons, which had been largely lacking prior to 1992 (Kempton et al., 1995, 2001; Davis, 1997; Markwick and Downes, 2000; Schmitz and Bowring, 2000, 2003a,b; Downes et al., 2002). These studies reveal a great diversity in lower-crustal lithologies within Archean cratons, which appear to correlate with seismic structure of the crust. Lower-crustal xenoliths from the Archean part of the Baltic (or Fennoscandian) Shield, like their post-Archean counterparts, are dominated by mafic lithologies (Kempton et al., 1995, 2001; Markwick and Downes, 2000; Ho¨ltta¨ et al., 2000). Most equilibrated at depths of 22–50 km and contain hydrous phases (amphibole  biotite). Partial melting and restite development is evident in some migmatitic xenoliths, but cumulates are absent (Kempton et al., 1995, 2001; Ho¨ltta¨ et al.,2000). A curious feature of these samples is the common occurrence of potassic phases (e.g., potassium feldspar, hornblende, biotite) in otherwise mafic granulites. These mafic xenoliths have been interpreted to represent gabbroic intrusions that underplated the Baltic Shield during the Paleoproterozoic flood-basalt event (2.4–2.5 Ga) and later experienced potassium-metasomatism coincident with partial melting at
1.8 Ga, a major period of granitic magmatism in this region (Kempton et al., 2001; Downes et al., 2002). The dominately mafic compositions of these xenoliths is consistent with the thick layer of high-velocity (7 km s1) material imaged beneath the Archean crust of the Baltic Shield (Luosto et al., 1989, 1990). The xenolith studies suggest that this layer formed during Paleoproterozoic basaltic underplating and is not part of the original Archean architecture of this Shield. In contrast to the Baltic Shield, mafic granulites appear to be absent in lower-crustal xenolith suites from the Archean Kaapvaal craton, which are dominated by metapelite and unique ultra-high-temperature granulites of uncertain petrogenesis (Dawson et al., 1997; Dawson and Smith, 1987;

26

Composition of the Continental Crust

2.0

2.0

Transition metals 1.5

1.5

1.0

1.0

0.5

0.5

3.1

High field-strength Mn

(a)

Sc

V

Cr

Co

Ni

Cu

Zn

2.0

(b)

Ti

Zr

Hf

Nb

Ta

Mo

2.0

Rare earth

2.6

1.5

1.5

1.0

1.0

0.5

0.5 Weaver and Tarney Shaw et al.

Alkali, alkaline earth and actinides Li

(c)

Rb

Cs

Sr

Ba

Pb

Th

U

(d)

La

Ce

Nd

Gao et al. Rudnick and Fountain

Sm

Eu

Gd

Yb

Lu

Figure 10 Comparison of the trace-element composition of the middle continental crust as determined by sampling of surface exposures (Shaw et al., 1994; Weaver and Tarney, 1984) and inferred from middle-crustal seismic velocities combined with surface and xenolith samples (Rudnick and Fountain, 1995; Gao et al., 1998a). All values normalized to the new composition provided in Table 5 (‘R&G’), which is an average of the values of Gao et al. (1998a) and Rudnick and Fountain (1995). Gray shaded field represents 20% variation from this value. (a) transition metals, (b) high-field strength elements, (c) alkali, alkaline earth and actinides, and (d) REEs. 1000

Middle crust 100

10

Chondrite normalized 1 La Ce Pr Nd

Sm Eu Gd Tb Dy Ho Er Tm Yb Lu

1000 Weaver and Tarney Shaw et al. Rudnick and Fountain

100

Gao et al. This study

10 Mantle normalized 1

Rb Cs

Ba

Th

K U

La Nb

Pb

Ce

Sr Pr

Zr Nd

Sm Hf

Eu

Ho

Ti Y

Yb

Figure 11 REE (upper) and multi-element plot (lower) of the compositions of the middle crust given in Table 5. Chondrite values from Taylor and McLennan (1985) and primitive mantle values from McDonough and Sun (1995).

Schmitz and Bowring, 2003a,b). These xenoliths derive from depths of > 30 km and show evidence for multiple thermal metamorphic overprints starting with ultrahigh temperature metamorphism at
2.7 Ga, which is associated with Ventersdorp magmatism (Schmitz and Bowring, 2003a,b). The absence of mafic granulites is consistent with the relatively low P-wave velocities in the lower crust of the Kaapvaal craton (Durrheim and Green, 1992; Nguuri et al., 2001; Niu and James, 2002), but it is not clear whether the lack of a mafic lower crust reflects the original crustal structure of this Archean craton (Nguuri et al., 2001) or reflects loss of a mafic complement some time after crust formation in the Archean (Niu and James, 2002). Lower crustal xenoliths from the Hannuoba basalts, situated in the central zone of the North China Craton, show a diversity of compositions ranging from felsic to mafic metaigneous granulites and metapelites (Gao et al., 2000; Chen et al., 2001; Liu et al., 2001; Zhou et al., 2002); approximately half the xenoliths have evolved compositions (Liu et al., 2001). All granulite xenoliths equilibrated under high temperatures (700–1000  C), corresponding to depths of 25–40 km (Chen et al., 2001), but mafic granulites yield higher temperatures than metapelitic xenoliths, suggesting their derivation from deeper crustal levels (Liu et al., 2001). Liu et al. used regional seismic refraction data and the lithologies observed in the Hannuoba xenoliths to infer the lower-crust composition in this part of the North China craton. They describe a layered lower crust in which the upper portion (from 24 km to
38 km, Vp
6.5 km s1), consists largely of felsic granulites and metasediments, and is underlain by a ‘lowermost’ crust (38–42 km, Vp
7.0 km s1) composed of intermediate granulites, mafic granulites, pyroxenite, and peridotite. Thus, the bulk lower crust in this region is intermediate in composition, consistent with the relatively large proportion of evolved

Table 6

Geochemical and mineral chemical studies of lower crustal xenoliths

Locality North America Nunivak Island, Alaska Central Slave Province, Canada Kirkland Lake, Ontario, Superior craton

Host

Xenolith types

Types of analyses

Pipe age

Crust age

Age

AB K

MG MGG, FG, MG

ME, Min. U–Pb

<5 Ma 50–70 Ma

Phanerozoic Archean

K

AN, MG

U–Pb

160 Ma

Archean

K K

MG, PG MG, PG, FG

Min. ME, TE, Min.


100 Ma <400 Ma

Proterozoic Phanerozoic

Evol. B K

PG

Sr, Nd, Pb

<2 Ma

Archean


2.8 Ga

Leeman et al. (1985, 1992)

MGG, MG, FG

ME, TE, O, U–Pb

45 Ma

Archean


2.6 Ga

AB

MG

FI

<1 Ma

Phanerozoic

Collerson et al. (1989), Kempton and Harmon (1992), Moecher et al. (1994), Rudnick et al. (1999) Ertan and Leeman (1999)

K AB

MGG, EC MGG, EC, FG, MP

<230 Ma 8–11 Ma

Proterozoic Proterozoic

180 Ma

K AB

MGG, MG MP, FG, IG, QZ, MGG

Min. ME, TE, Min., Sr, Nd, O, U–Pb ME, Min. ME, TE, Min., Sr, Nd, Pb, U–Pb ME, TE, Min., Sr, Nd ME, TE, Min., Sr, Nd, Pb ME, TE, Min.

Proterozoic Archean


1.7 Ga

25–30 Ma

Proterozoic


1.8 Ga

23–27 Ma

Proterozoic

1.2–1.9 Ga

Ehrenberg and Griffin (1979), Broadhurst (1986), Wendlandt et al. (1993, 1996), Mattie et al. (1997), Condie et al. (1999) Esperanca et al. (1988)

25 Ma

Proterozoic

<20 Ma

Proterozoic


1.9 Ga

Arculus and Smith (1979), Schulze and Helmstadt (1979), Arculus et al. (1988) Chen and Arculus (1995)

<3 Ma

Proterozoic

1.1–1.4 Ga

Kempton et al. (1990), Kempton and Harmon (1992)

<1 Ma

Proterozoic

1.5 Ga

2.5 Ga, 1.3 Ga (meta) 2.6–2.8 Ga, 2.4–2.5 Ga (meta)

References

Francis (1976) Davis (1997) Moser and Heaman (1997)

Ayer’s cliff, Quebec Popes Harboura, Nova Scotia Snake River Plainsa, Idaho Bearpaw Mts., Montana Simcoe Volcanic Field, Washington Cascades Riley County, Kansas Central Sierra Nevada, California Colorado/Wyoming Mojave Desert, California Navajo Volcanic Field, Colorado Plateau Camp Creek, Ariszona

K

MGG, FG, AM, EC, MP

L

MGG, AM, EC

Chino Valley, Arizona

L

MGG, AM

San Francisco Volcanic Field Geronimo Volcanic Field, New Mexico Kilbourne Hole, New Mexico

AB

MG, IG

AB

MG, IG

AB

MG, AN, PG, FG

AB

MG

Min., ME

<3 Ma

Proterozoic

Padovani and Carter (1977), Davis and Grew (1977), James et al. (1980), Padovani et al. (1982), Reid et al. (1989), Leeman et al. (1992), Scherer et al. (1997). Baldridge (1979)

AB

MG

Min.

<3 Ma

Proterozoic

Warren et al. (1979)

AB

MGG, IG, FG

ME, Sr, Nd, Pb, U– Pb

<40 Ma

Proterozoic

1.1 Ga

Meyer and Brookins (1976) Dodge et al. (1986, 1988), Domenick et al. (1983), Ducea and Saleeby (1996, 1998) Bradley and McCallum (1984) Hanchar et al. (1994)

Cameron and Ward (1998)

27

(Continued)

Composition of the Continental Crust

Elephant Butte, New Mexico Engle Basin, New Mexico West Texas

ME, TE, Min., Sr, Nd ME, TE, Min., Sr, Nd, Pb, O Min., ME, Sr, Nd, Pb, O, U–Pb

Trzcienski and Marchildon (1989) Owen (1988), Eberz (1991)

28

Table 6

(Continued) Host

Xenolith types

Types of analyses

Pipe age

Crust age

Age

References

Northern Mexico

AB

MG, PG, FG

ME, TE, Min., Sr, Nd, C, U–Pb

<25 Ma

Proterozoic

From 1 to 1400 Ma

Central Mexico San Luis Potosi, Central Mexico South America Mercaderes, SW Columbia Salta Rift, NW Argentina Calbuco Volcano, Chile

AB AB

MG MG, MGG, IG

ME, Min, Nd ME, TE, Min., Sr, Nd

<1 Ma

Phanerozoic Proterozoic

1.5 Ga
1.2 Ga

Nimz et al. (1986), Ruiz et al. (1988a,b), Roberts and Ruiz (1989), Hayob et al. (1989), Rudnick and Cameron (1991), Cameron et al. (1992), Moecher et al. (1994), Smith et al. (1996), Scherer et al. (1997) Urrutia-Fucagauchi and Uribe-Cifuentes (1999) Schaaf et al. (1994)

AB

MG, MGG, IG, HB,

<10 Ma

Phanerozoic

AB

FG, MG

Mesozoic

Proterozoic

AND

MG

<1000 years

Paleozoic

Hickey-Vargas et al. (1995)

Pali Aike, Southern Chile Europe Scotland, Northern Uplands

AB

MG

ME, TE, Min., Sr, Nd, Pb Min., ME, TE, Sr, Nd, Pb ME, TE, Min., Sr, Nd ME, Min., FI

<3 Ma

Phanerozoic

Selverstone and Stern (1983)

AB

MG, AN, IG, MP, MGG, HB

ME, TE, Min., Nd, Sr, U–Pb


300 Ma

Archean/ Proterozoic

360 Ma, 1.8 Ga

Eastern Finland

K

MG, MGG, AM

525 Ma

Archean

1.7–2.6 Ga

Arkhangelsk Kimberlite, Baltic shield, Russia Elovy island, Baltic shield, Russia Belarus, Russia

K

MGG

ME, TE, Min., Nd, U–Pb ME, TE, Min., Sr, Nd

van Breeman and Hawkesworth (1980), Upton et al. (1983), Halliday et al. (1984), Hunter et al. (1984), Upton et al. (1998, 2001) Ho¨ltta¨ et al. (2000)

360 Ma

Archean

1.7–1.9 Ga

Markwick and Downes (2000)

K

MGG, EC, FG, AM

360–380 Ma

Archean


1.8 Ga, 2.4– 2.5 Ga

Kempton et al. (1995, 2001), Downes et al. (2002)

K

MGG, EC, HB

AB

MG, MGG

AB

MGG, AM

ME, TE, Min., Sr, Nd, Pb, U–Pb ME, TE, Min., Sr, Nd ME, TE, Min., Sr, Nd, O ME, TE, Min., Sr, Nd, Hf, Pb, O

AB

AM

Pannonian basin, W. Hungary Kampernich, E. Eifel, Germany

Weber et al. (2002)
1.8 Ga

370 Ma

Lucassen et al. (1999)

Markwick et al. (2001)

2–5 Ma

Phanerozoic

<1 Ma

Phanerozoic

ME, TE, Min.

<1 Ma

Phanerozoic

Wehr Volcanoa, E. Eifel, Germany N. Hessian Depression, Germany Massif Central, France

AB

MGG, MG, PG, FG

ME, TE, Min., O

<50 Ma

Phanerozoic

AB

MGG, MG, PG, FG

ME, TE, Nd, Sr, Pb, O, C

<5 Ma

Phanerozoic

Tallante, Spain Central Spain

AB K

PG IG, FG, MP

Min. ME, TE, Min.

<20 Ma Early Mesozoic

Phanerozoic Proterozoic

1.5 Ga or
450 Ma?

Embey-Isztin et al. (1990), Kempton et al. (1997), Embey-Isztin et al. (2003), Dobosi et al. (2003) Okrusch et al. (1979), Stosch and Lugmair (1984), Rudnick and Goldstein (1990), Loock et al. (1990), Kempton and Harmon (1992), Sachs and Hansteen (2000) Wo¨rner et al. (1982), Grapes (1986) Mengel and Wedepohl (1983) Mengel (1990)


350 Ma

Leyreloup et al. (1977), Dostal et al. (1980), Vidal and Postaire (1985), Downes and Leyreloup (1986), Kempton and Harmon (1992), Downes et al. (1991), Moecher et al. (1994) Vielzeuf (1983) Villaseca et al. (1999) (Continued)

Composition of the Continental Crust

Locality

AB

MG

Min.


3 Ma

Phanerozoic

Rutter (1987)

AB K

MG, AN, PG MGG, AN, EC

ME, TE ME, Min., U–Pb

<20 Ma 90–120 Ma

Proterozoic Archean

Leyreloup et al. (1982) Toft et al. (1989), Barth et al. (2002)

AB

MGG, EC, AN

<20 Ma

Proterozoic

AB K

MGG, MG PG

ME, TE, Min., Sr, Nd, Pb, C Min., ME, TE Min., U–Pb

<3 Ma 90–140 Ma

Proterozoic Archean


2.7 Ga (meta)

Dawson (1977), Jones et al. (1983), Cohen et al. (1984), Moecher et al. (1994) Thomas and Nixon (1987) Dawson and Smith (1987), Dawson et al. (1997), Schmitz and Bowring (2003a,b)

K

PG

U–Pb

114 Ma

Archean


2.7 Ga

Schmitz and Bowring (2003a,b)

K

MGG, MG, FG, MP, EC

ME, TE, Min., Sr, Nd, Pb, U–Pb

90–140 Ma

Proterozoic

Orapa, Zimbabwe Craton, Botswana

K

MP

U–Pb

93 Ma

Proterozoic

Davis (1977), Rogers and Hawkesworth (1982), Griffin et al., 1979), Rogers (1977), van Calsteren et al. (1986), Huang et al. (1995), Schmitz and Bowring (2003a) Schmitz and Bowring (2003a)

Central Cape Province, Eastern Namaqualand, South Africa

K

MGG, AM

ME, TE, Min., Sr, Nd, Pb, U–Pb

90–140 Ma

Proterozoic

1.4 Ga, 1.1–1.0 Ga (meta) 2.0 Ga (meta), 1.24 Ga (meta)
1.1 Ga (meta)

AB AB AB AB

MGG MGG, Am MG MG

Min. TE Min.


100 Ma 10 000 years <2 Ma

Phanerozoic Phanerozoic Proterozoic

Esperanca and Garfunkel (1986), Mittlefehldt (1986) Mittlefehldt (1984) Nasir (1992), Nasir (1995) Nasir and Safarjalani (2000)

K

MGG

Min.

Archean

Shatsky et al. (1990, 1983)

AB

MG, MGG, IG, AM

<5 Ma

Proterozoic

AB

14–27 Ma

Archean

AB

FG, IG, MG, MGG, AN, MP MGG

ME, TE, Min., Sr, Nd, Pb, O ME, TE, Sr, Nd, Pb, U–Pb ME, TE, Min.

Mesozoic

Archean

Kempton and Harmon (1992), Kopylova et al. (1995), Stosch et al. (1995) Gao et al. (2000), Liu et al. (2001), Chen et al. (2001), Zhou et al. (2002) Zheng et al. (2003)

AB

MG

AB

MG, MGG, IG, FG

Ichinomegata, Japan

AND

AM

ME, TE, Min., Sr, Nd ME, TE, Min., Sr

Deccan Traps, India

K

MG

ME, Min.

Sardinia, Italy Africa Hoggar, Algeria Man Shield, Sierra Leone Lashaine, Tanzania Fort Portal, Uganda Free State Kimberlites, Kaapvaal Craton, South Africa Newlands Kimberlite, Kaapvaal Craton, South Africa Lesotho, South Africa

Middle East Mt. Carmel, Israel Birket Ram, Israel Jordan Shamah volcanic fields, Syria


2.5, 1.9, 0.4, 0.22 Ga

Lee et al. (1993) <20 Ma 10 000 years

Archean/ Proterozoic Phanerozoic

Yu et al. (2003)

(Continued)

29

Kuno (1967), Aoki (1971), Zashu et al. (1980), Tanaka and Aoki (1981) Dessai et al. (1999), Dessai and Vasseli (1999)

Composition of the Continental Crust

Asia Udachnaya, Siberia, Russia Tariat Depression, Central Mongolia Hannuoba, North China Craton Xinyang, North China craton Penghu Islands, SE China Southeastern China

van Calsteren et al. (1986), Pearson et al. (1995), Schmitz and Bowring (2000), Schmitz and Bowring (2003a)

(Continued)

Locality

30

Table 6

Host

Xenolith types

Pipe age

Crust age

Min.

Australia–New Zealand–Antarctica McBride Province, N. AB MGG, MG, PG, FG Queensland

ME, TE, Min.,Sr, Nd, Pb, O, C, U–Pb

<3 Ma

Proterozoic

300 Ma and
1.6 Ga <100 Ma

AB

MGG, MG

ME, TE, Min., Sr, Nd, Pb, O

<1 Ma

Phanerozoic

Central Queensland

AB

MG, MGG

50 Ma

Phanerozoic

Gloucester, NSW Sydney Basin Boomi Creek, NSW Delegate, NSW

AB AB AB AB

MGG, MG MG MG MG, MGG, FG, EC

50 Ma 50 Ma 50 Ma
140 Ma

Phanerozoic Phanerozoic Phanerozoic Phanerozoic

Jugiong, NSW White Cliffs, NSW Anakies, Victoria

K K AB

MG, MGG MGG MG, MGG

<17 Ma
260 Ma <2 Ma

Phanerozoic Proterozoic Phanerozoic

El Alamein, South Australia Calcutteroo, South Australia

K

MGG, EC

ME, TE, Min., Sr, Nd ME, TE, Sr, Nd ME, TE, Sr, Nd ME, TE, Min. ME, TE, Min., Sr, Nd, U–Pb Min. Min. ME, TE, Min., Sr, Nd ME, TE, Min.


170 Ma

Proterozoic

K

MGG, FG, EC

ME, TE, Min., Sr, Nd, U–Pb


170 Ma

Proterozoic

AB

MG

ME, TE, Min

AB

MG, MGG

ME, TE, Min.,Sr, O

AB

MG, SG

ME, Min.

Indian Ocean Kergulen Archipelago a

References Hacker et al. (2000)

Chudleigh Province, N. Queensland

Banks Penninsula, New Zealand Mt. Erebus Volcanic Field,

Age

400 Ma

1.6–1.5 Ga, 780 Ma, 620 Ma, 330 Ma

Kay and Kay (1983), Rudnick and Taylor (1987), Rudnick and Williams (1987), Stolz and Davies (1989), Stolz (1987), Rudnick (1990), Rudnick and Goldstein (1990), Kempton and Harmon (1992), Moecher et al. (1994) Kay and Kay (1983), Rudnick et al. (1986), Rudnick and Taylor (1991), O’Reilly et al. (1988), Rudnick and Goldstein (1990), Kempton and Harmon (1992) Griffin et al. (1987), O’Reilly et al. (1988) Griffin et al. (1986), O’Reilly et al. (1988) Griffin et al. (1986), O’Reilly et al. (1988) Wilkinson (1975), Wilkinson and Taylor (1980) Lovering and White, (1964, 1969), Griffin and O’Reilly (1986), O’Reilly et al. (1988), Arculus et al. (1988), Chen et al. (1998) Arculus et al. (1988) Arculus et al. (1988) Sutherland and Hollis (1982), Wass and Hollis (1983), O’Reilly et al. (1988) Edwards et al. (1979), Arculus et al. (1988) McCulloch et al. (1982), Arculus et al. (1988), Chen et al. (1994)

Sewell et al. (1993) <5 Ma

Phanerozoic/ Proterozoic

Kyle et al. (1987), Kalamarides et al. (1987),Berg et al. (1989)

Phanerozoic/ Proterozoic

McBirney and Aoki (1973), Gregoire et al. (1998, 1994)

Xenoliths from these localities are probably derived from mid-crustal levels based on either: equilibration pressures, lack of mantle derived xenoliths in the same hosts and/or chemically evolved character of the host. Only papers in which data are reported are listed here. Abbreviations Host types: AB ¼ alkali basaltic association; AND ¼ andesite; K ¼ kimberlitic association (including lamproites, minettes, kimberlites), Evol. B. ¼ evolved basalt; L ¼ latite. Xenolith types: AM ¼ amphibolite; AN ¼ anorthosite; EC ¼ elcogite; FG ¼ felsic granuilte; HB ¼ hornblendite; IG ¼ intermediate granulite; MG ¼ mafic granulite; MGG ¼ mafic garnet granulite; MP ¼ metapelite, PG ¼ paragneiss, QZ ¼ quartzite, SG ¼ saphirine granulites. Types of analyses: FI ¼ fluid inclusions, ME ¼ major element analyses; Min ¼ mineral analyses; TE ¼ trace element analyses; Sr ¼ Sr isotope analyses; Nd ¼ Nd isotope analyses; Pb ¼ Pb isotope analyses, O ¼ oxygen isotope analyses, C ¼ carbon isotope analyses, U–Pb ¼ U–Pb geochronology on accessory phases (zircon, rutile, titanite, etc.).

Composition of the Continental Crust

Tibetan plateau

Types of analyses

Composition of the Continental Crust

granulites at Hannuoba. Zircon geochronology shows that mafic granulites and some intermediate granulites were formed by basaltic underplating in the Cretaceous. This mafic magmatism intruded pre-existing Precambrian crust consisting of metapelites that had experienced high-grade metamorphism at 1.9 Ga (Liu et al. (2001) and references therein). Fragmentary xenolithic evidence for the composition of the lower crust is available for three other Archean cratons. Two mafic garnet granulites from the Udachnaya kimberlite in the Siberian craton yield Archean lead–lead and Proterozoic samarium–neodymium mineral isochrons (Shatsky, Rudnick and Jagoutz, unpublished data). It is likely that that lead–lead isochrons are frozen isochrons yielding anomalously old ages due to ancient uranium loss; the best estimate of the true age of these mafic granulites is Proterozoic. Moser and Heaman (1997) report Archean uranium–lead ages for zircons derived from mafic lower-crustal xenoliths from the Superior Province, Canada. They suggest these samples represent the mafic lower crust presently imaged seismically beneath the Abitibi greenstone belt, but which is not exposed in the Kapuskasing uplift. These granulites experienced an episode of high-grade metamorphism at 2.4 Ga, which Moser and Heaman (1997) attribute to underplating of basaltic magmas associated with the opening of the Matachewan Ocean. Davis (1997) reports mafic to felsic granulite xenoliths from the Slave craton, Canada, that have Archean to Proterozoic uranium–lead zircon ages. The mafic granulites appear to derive from basaltic magmas that underplated the felsicArchean crust during the intrusion of the 1.3 Ga McKenzie dike swarm. The above case studies illustrate the utility of lower crustal xenolith studies in defining the age, lithology, and composition of the lower crust beneath Archean cratons. When viewed collectively, an interesting generality emerges: when mafic granulites occur within the lower crust of Archean cratons they are generally inferred to have formed from basaltic underplating related to post-Archean magmatic events (In addition to the studies mentioned above is the case of the thick, highvelocity lower crust beneath the Archean Wyoming Province and Medicine Hat Block, western North America, which is also inferred to have formed by Proterozoic underplating based on uranium–lead zircon ages from lower crustal xenoliths (Gorman et al., 2002)). Only granulites from the Superior Province appear to represent Archean mafic lower crust (Moser and Heaman, 1997). This generality is based on still only a handful of studies of xenoliths from Archean cratons and additional studies are clearly needed. However, if this generality proves robust, it implies that the processes responsible for generation of crust in most Archean cratons did not leave behind a mafic lower crust, the latter of which is commonly observed in post-Archean regions (Rudnick (1992) and references therein). It may be that this mafic lower crust was never produced, or that it formed but was removed from the crust, perhaps via density foundering (R. W. Kay and S. M. Kay, 1991; Gao et al., 1998b; Jull and Kelemen, 2001, see also Chapter 4.12). In either case, the apparent contrast in lowercrustal composition between Archean and post-Archean regions, originally pointed out by Durrheim and Mooney (1994), suggests different processes may have been operative in the formation of Archean crust (see Chapter 4.11). We

31

return to the issue of what crust composition tells us about crustal generation processes in Section 4.1.5. In summary, despite the uncertainties regarding the representativeness of any given lower-crustal xenolith suite (Rudnick, 1992), the above studies show that an accurate picture of the deep crust can be derived from such studies, especially when xenolith studies are combined with seismological observations of lower-crust velocities, to which we now turn.

4.1.3.5.2 Seismological evidence The P-wave velocity of the lower crust varies from region to region, but average, temperature-corrected velocities for lower crust from a variety of different tectonic settings are high (6.9–7.2 km s1; Rudnick and Fountain, 1995; Christensen and Mooney, 1995). Such velocities are consistent with the dominance of mafic lithologies (mafic granulite and/or amphibolite) in these lower-crustal sections. High-grade metapelite, in which much of the quartz and feldspars have been removed by partial melting, is also characterized by high seismic velocities and thus may also be present (Rudnick and Fountain, 1995). Although seismically indistinct, some limit on the amount of metapelite in these high-velocity layers can be made on the basis of heat-flow and xenolith studies; these suggest that metapelite is probably a minor constituent of the lower crust (i.e., <10%; Rudnick and Fountain, 1995). In addition, average P-wave velocities for mafic granulite or amphibolite are higher than those observed in many lower-crustal sections (corrected to roomtemperature velocities). Average room-temperature P-wave velocities for a variety of mafic lower crustal rock types are generally equal to or higher than 7 km s17.0  0.2 km s1 for amphibolite, 7.0 to 7.2  0.2 km s1 for garnet-free mafic granulites, and 7.2 to 7.3  0.2 km s1 for garnet-bearing mafic granulites at 600 MPa (Rudnick and Fountain, 1995; Christensen and Mooney, 1995). Lower-crustal sections having temperature-corrected P-wave velocities of 6.9–7.0 km s1 (e.g., Paleozoic orogens and Mesozoic/Cenozoic extensional and contractional terranes), are thus likely to have lower-velocity rock types present (up to 30% intermediate to felsic granulites), in addition to mafic granulites or amphibolites (Rudnick and Fountain, 1995). Although the average lower-crustal seismic sections discussed above show high velocities, some sections are characterized by much lower velocities, indicating a significantly more evolved lower-crust composition. For example, the crust of a number of Archean cratons is relatively thin (
35 km) with low seismic velocities in the lower crust (6.5–6.7 km s1), suggesting an evolved composition (e.g., Yilgarn craton (Drummond, 1988), Kaapvaal craton (Durrheim and Green, 1992; Niu and James, 2002), and North China craton (Gao et al., 1998a,b)). As discussed above, it is not clear whether these thin and relatively evolved regions of Archean crust represent the original crustal architecture, formed by processes distinct from those responsible for thicker and more mafic crustal regions (e.g., Nguuri et al., 2001), or reflect loss of a mafic layer from the base of the original crust (Gao et al., 1998b; Niu and James, 2002). In addition, some Cenozoic– Mesozoic extensional and contractional regions, Paleozoic orogens, and active rifts show relatively slow lower-crustal velocities of 6.7–6.8 km s1 and may contain >40% felsic and intermediate granulites (Rudnick and Fountain, 1995). Two extreme

32

Composition of the Continental Crust

examples are the southern Sierra Nevada and Central Andean backarc. In both cases, the entire crustal columns are characterized by P-wave velocities of less than 6.4 km s1 (Beck and Zandt, 2002; Wernicke et al., 1996). A relatively high-velocity (Vp¼ 6.4–6.8 km s1) layer of <5 km in thickness occurs only at the base of the Central Andean backarc at
60 km depth. In summary, the seismic velocity of the lower crust is variable from region to region, but is generally high, suggesting a dominance of mafic lithologies. However, most seismic sections require the presence of evolved compositions in addition to mafic lithologies in the lower crust (up to 30% for average velocity of 6.9 km s1) and a few regions (e.g., continental arcs and some Archean cratons) are characterized by slow lower crust, indicating a highly evolved average composition. This diversity of lithologies is consistent with that seen in both crustal cross-sections and lower-crustal xenolith suites and also provides a mechanism (lithological layering) to explain the common occurrence of seismic reflections observed in many seismic reflection profiles (Mooney and Meissner, 1992).

4.1.3.5.3 Lower-crust composition Table 7 lists previous estimates of the composition of the lower crust. These estimates include averages of exposed granulites (columns 1 and 2; Weaver and Tarney, 1984; Shaw et al.,1994), averages of individual lower-crustal xenolith suites (columns 3–6, Condie and Selverstone, 1999; Liu et al., 2001; Rudnick and Taylor, 1987; Villaseca et al., 1999), the median composition of lower-crustal xenoliths (column 7; updated from Rudnick and Presper (1990), with data from papers cited in Table 6 (the complete geochemical database for lower crustal xenoliths is available on the GERM web site http://earthref.org/ cgi-bin/erda.cgi?n¼1,2,3,8 and also on the Treatise web site), averages derived from linking seismic velocity data for the lower crust with the compositions of lower-crustal rock types (columns 8–10; Rudnick and Fountain, 1995; Wedepohl, 1995; Gao et al., 1998a), and Taylor and McLennan’s model lower crust (column 11). It is readily apparent from this table and Figures 12 and 13 that, compared to estimates of the upper-crust composition (Table 1), there is much greater variability in estimates of the lower-crust composition. For example, TiO2, MgO, FeOT, and Na2O all vary by over a factor of 2, CaO varies by almost a factor of 7, and K2O varies by over an order of magnitude between the different estimates (Figures 12 and 13). Trace elements show correspondingly large variations (Figure 13). In contrast, modern estimates of major elements in the upper crust generally fall within 20% of each other (Table 1 and Figure 2 – gray shading) and most trace elements fall with 50%. We now explore the possible reasons for these variations, with an eye towards determining a ‘best estimate’ of the global lower-crust composition. The two lower-crustal estimates derived from averages of surface granulites are generally more evolved than other estimates (Table 7, and Figures 12 and 13). Weaver and Tarney’s (1984) lower-crustal estimate derives from the average of Archean Scourian granulites in the Lewisian complex, Scotland. It is one of the most evolved compositions given in the table and is characterized by a steeply fractionated REE pattern, which is characteristic of Archean granitoids of the tonalite–trondhjemite–granodiorite assemblage (see Chapter 4.11) and severe depletions in the large-ion

lithophile elements, in addition to thorium and uranium (Rudnick et al., 1985). The estimate of Shaw et al. (1994) derives from a weighted average of the granulite-facies rocks of the Kapuskasing Structural Zone, Canadian Shield. The average is intermediate in overall composition. As discussed in Section 4.1.3.4.1, the Kapuskasing cross-section provides samples down to depths of
25 km, leaving the lower 20 km of lower crust unexposed. Seismic-velocity data show this unexposed deepest crust to be mafic in bulk composition, consistent with the limited data for lower-crustal xenoliths from the Superior province (Moser and Heaman, 1997). Thus, the lower-crustal estimates of Weaver and Tarney (1984) and Shaw et al. (1994) may be representative of evolved lower crust in Archean cratons lacking a high-velocity lower crust, but are unlikely to be representative of the global continental lower crust (Archean cratons constitute only
7% of the total area of the continental crust (Goodwin, 1991)). The lower-crustal estimates derived from particular xenolith suites (columns 3–6, Table 7) were selected to illustrate the great compositional heterogeneity in the deep crust. The average weighted composition of lower-crustal xenoliths from central Spain (Villaseca et al., 1999) is one of the most felsic compositions in Table 7 (with
63 wt% SiO2, Figure 12(b)). It has higher K2O content than nearly every estimate of the upper crust composition (Table 1), and has such a high heat production (0.8 mW m3), that a 40 km thickness of crust with average upper- and middle-crustal compositions given in Tables 3 and 5, respectively, would generate a surface heat flow of 41 mW m2. This is equivalent to 100% of surface heat flow through Archean crust, 85% of surface heat flow through Proterozoic crust, and 71% of the surface heat flow through Paleozoic crust (see Chapter 4.2). Assuming the heat flux through the Moho is
17 mW m2 (see Chapter 4.2) this lower-crust composition could thus be representative of the lower crust in Phanerozoic regions with high surface heat flow, but clearly cannot be representative of the global average lower crust. Likewise, the ‘evolved’ xenolith-derived lower crustal estimate of Liu et al. (2001) for the central zone of the North China craton, also has a high-K2O content that exceeds that in most modern estimates of the upper continental crust (Table 1 and Figure 12(b)). Total crustal heat production calculated as described above using the Liu et al. composition for the lower crust yields a value of 0.95 mW m3, which corresponds to a surface heat flow of 34 mW m2. This composition is thus unlikely to be representative of the lower crust in Archean cratons, where average surface heat flow is 41  1 mW m2 (Nyblade and Pollack, 1993). However, surface heat flow through this part of the North China craton is unusually high (50 mW m2; Hu et al., 2000), thus permitting a more radiogenic lower crust in this region. Other peculiarities of this bulk composition include an extreme Mg# of 76 (Mg# ¼ 100 * molar Mg/(Mg þ Fe)) and extreme nickel (347 ppm) and chromium (490 ppm) contents. These values are especially unusual given the rather felsic bulk composition of this estimate and reflect the very high Mg# mafic granulites present in this suite and the inclusion of up to 25% peridotite within the lower-crustal mixture modeled by Liu et al. (2001). This estimate also has the highest strontium and barium contents of all estimates (712 ppm and 1434 ppm, respectively) and is the most HREE depleted (Table 7). The two remaining average lower-crustal xenolith suites (Rudnick and Taylor, 1987

Compositional estimates of the lower continental crust. Major elements in weight percent. Trace element concentration units the same as in Table 2

Table 7

2 Shaw et al. (1994)

3 Rudnick and Taylor (1987)

4 Condie and Selverstone (1999)

5 Villaseca et al. (1999)

6 Liu et al. (2001)

7 Global Lower Crustal Xenolith

8 Rudnick and Fountain (1995)

9 Wedepohl (1995)

10 Gao et al. (1998a)

11 Taylor and McLennan (1985, 1995)

62.9 0.5 16.0 5.4 0.08 3.5 5.8 4.5 1.0 0.19 53.4

58.3 0.65 17.4 7.09 0.12 4.36 7.68 2.70 1.47 0.24 52.3 14

49.6 1.33 16.4 12.0 0.22 8.72 10.1 1.43 0.17

52.6 0.95 16.4 10.5 0.16 6.04 8.50 3.19 1.37 0.21 50.5

62.7 1.04 17.4 7.52 0.10 3.53 1.58 2.58 3.41 0.16 45.6

59.6 0.60 13.9 5.44 0.08 9.79 4.64 2.60 3.30 0.13 76.2 3.3

52.0 1.13 17.0 9.08 0.15 7.21 10.28 2.61 0.54 0.13 58.6 5

53.4 0.82 16.9 8.57 0.10 7.24 9.59 2.65 0.61 0.10 60.1 6

59.0 0.85 15.8 7.47 0.12 5.32 6.92 2.91 1.61

59.8 1.04 14.0 9.30 0.16 4.46 6.20 3.00 1.75 0.21 46.1 13 1.1 7.6

54.3 0.97b 16.1 10.6 0.22 6.28 8.48 2.79 0.64b

56.5

3.2

88 58

11 569 7 202 5

16 140 168 38 75 28 83

33 217 276 31 141 29

41 447 16 114 5.6

12 196 28 127 13 0.8

28 133 20 73

37 518 86 7.75

17 139 178 22 65 40 83

20 100 490 31 347

90 286 40 206 15

51 712 8 180 6.4

89 15

29 189 145 41 80 32 85 17

31 196 215 38 88 26 78 13

7 354 20 68 5.6 0.8

11 348 16 68 5.0

55.9 13 1.7 5 34 429 408 278 25 149 228 38 99 37 79 17 1.4 1.3 0.17 0.28 41 352 27 165 11 0.6

703 231 216 26 185 123 36 64 50 102 19 1.24 1.6 0.17

35b 271b 219b 36b 156b 90 83 18 1.6 0.8 0.05

56 308 18 162 10 0.54

12b 230 19 70 6.7b 0.8

2.78 51 0.097

1 90 0.098 0.050

33

80 0.101 0.052

51.4 11 1.0 8.3

Composition of the Continental Crust

SiO2 TiO2 A12O3 FeOTa MnO MgO CaO Na2O K2O P2O5 Mg# Li Be B N F S Cl Sc V Cr Co Ni Cu Zn Ga Ge As Se Br Rb Sr Y Zr Nb Mo Ru Pd Ag Cd In

1 Weaver and Tarney (1984)

(Continued)

(Continued)

Table 7

Pb Bi Th U a

3 Rudnick and Taylor (1987)

4 Condie and Selverstone (1999)

5 Villaseca et al. (1999)

6 Liu et al. (2001)

7 Global Lower Crustal Xenolith

8 Rudnick and Fountain (1995)

1.3

757 22 44

0.67 523 21 45

19 3.3 1.18

23 4.1 1.18

0.43

0.28

0.19 1.2 0.18 3.6

1.13 0.2 2.8 1.3

0.07 212 12 28 3.6 16 4.1 1.36 4.31 0.79 5.05 1.12 3.25 3.19 3.3

564 22 46

994 38 73

24 5.17 1.30 4.67 0.72

30 6.6 1.8 6.8

0.15 1434 18 36 14 2.59 0.97 0.33

6.7

2.09 0.37 1.9 0.5

4.0 0.65 2.1

0.79 0.12 4.6 0.3

0.5

13

6

3.3

9.8

0.42 0.05

2.6 0.66

0.54 0.21

1.64 1.38

5.74 0.47

0.19 305 9.5 21 [2.1] 13.3 3.40 1.20 3.6 0.50 3.9 0.6 2.0

0.3 259 8 20

1.70 0.30 1.9 0.5 0.5

1.5 0.25 1.9 0.6

11 2.8 1.1 3.1 0.48 3.1 0.68 1.9

12.9

4.1

4

0.49 0.18

0.50 0.18

1.2 0.2

9 Wedepohl (1995)

10 Gao et al. (1998a)

11 Taylor and McLennan (1985, 1995)

2.1 0.30 0.14 0.8 568 27 53 7.4 28 6.0 1.6 5.4 0.81 4.7 0.99

1.34 0.09

1.5 0.2

2.6 509 29 53

0.47b 150 11 23 2.8 13 3.17 1.17 3.13 0.59 3.6 0.77 2.2 0.32 2.2 0.29 2.1 0.7b 0.6b 0.4 0.05 0.13

25 4.65 1.39 0.86

2.5 0.43 4.0 0.8 0.6

2.29 0.38 4.2 0.6 0.51

0.021 0.26

2.87 1.58 0.0063 0.38

12.5 0.037 6.6 0.93

13 0.38 5.23 0.86

3.4 0.23 5.0b 0.038 2.0b 0.53b

Total Fe as FeO. Value from McLennan (2001b). 1.Weighted average of Scourian granulites, Scotland, from Weaver and Tarney (1984). 2. Weighted average of Kapuskasing Structural Zone granulites, from Shaw et al. (1994). 3. Average lower crustal xenoliths from the McBride Province, Queensland, Australia from Rudnick and Taylor (1987). 4. Average lower crustal xenoliths from the four corners region, Colorado Plateau, USA from Condie and Selverstone (1999). 5. Weighted mean composition calculated from lithologic proportions of lower crustal xenoliths from Central Spain from Villaseca et al. (1999). 6. Weighted average of lower crustal xenoliths from Hannuoba according to seismic crustal model of North China Craton from Liu et al. (2001). 7. Median worldwide lower crustal xenoliths from Rudnick and Presper (1990), updated with data from more recent publications. Complete database available at http://earthref.org/cgi-bin/ erda.cgi?n ¼ 1, 2, 3, 8, and on Treatise website. 8. Average lower crust derived from global average seismic velocities and granulites from Rudnick and Fountain (1995). 9. Average lower crust in western Europe derived from seismic data and granulite xenolith compositions from Wedepohl (1995). 10. Average lower crust derived from seismic velocities and granulite data from the North China craton from Gao et al. (1998a). 11. Average lower crust from Taylor and McLennan (1985, 1995), updated by McLennan and Taylor (1996) and McLennan (2001b). Mg# ¼ molar 100  Mg/(Mg þ Fetot). Values in [ ] are interpolated. Major elements are recast to 100% anhydrous. b

Composition of the Continental Crust

Sn Sb I Cs Ba La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu Hf Ta W Re Os Ir Pt Au Hg T1

2 Shaw et al. (1994)

34

1 Weaver and Tarney (1984)

Composition of the Continental Crust

2.0

Normalized to R and F

Terrains and models 1.5

1.0 Weaver and Tarney

0.5

Shaw et al. Gao et al. Wedepohl Taylor and McLennan

0.0 Si

(a)

Al

Fe

Mg

Ca

Na

K

Ca

Na

K

2.0

Normalized to R and F

Xenolith suites 1.5

1.0 N. Queensland

0.5

Colorado Plateau Central Spain E. China Median Xenolith

0.0 (b)

Si

Al

Fe

Mg

Figure 12 Comparison of different major-element estimates of the composition of the lower continental crust. All data normalized to the lower-crust composition of Rudnick and Fountain (1995), which is adopted here. Gray shaded field represents 10% variation from the model of Rudnick and Fountain (1995). (a) Models based on granulite terrains (Scourian granulites: Weaver and Tarney, 1984; Kapuskasing Structural Zone: Shaw et al., 1986), seismological models (Eastern China: Gao et al., 1998a,b; western Europe: Wedepohl, 1995) and Taylor and McLennan (1985, 1995; modified by McLennan, 2001b) model lower crust. (b) Models based on weighted averages of lower crustal xenoliths. These include: Northern Queensland, Australia (Rudnick and Taylor, 1987); Colorado Plateau, USA (Condie and Selverstone, 1999); Central Spain (Villaseca et al., 1999); eastern China (Liu et al., 2001) and the median global lower crustal xenolith composition, updated from Rudnick and Presper (1990). Note that K data for eastern China and Central Spain are co-incident on this plot, making it hard to distinguish the separate lines.

and Condie and Selverstone, 1999) both have mafic compositions that more closely approximate the global average-xenolith composition (column 7 in Table 7). It has been shown repeatedly from numerous xenolith studies that the majority of lower-crustal xenoliths are mafic in composition (Rudnick and Presper (1990), Rudnick (1992), and Downes (1993) and references therein). Thus, the ‘best estimate’ of the lower crust made on the basis of xenolith studies is found in column 7 of Table 7, which gives the median composition of all analyzed lower-crustal xenoliths. Yet it remains unclear to what degree xenolith compositions reflect average lower crust. Uncertainties include the degree to which volcanic pipes sample a representative cross-section of the deep crust and whether certain xenoliths (e.g., felsic xenoliths and metacarbonates) suffer preferential disaggregation or dissolution in the host magmas. In addition, large compositional variations are apparent from place to place (Figure 12(b)) and xenolith data are only available for limited regions of the continents.

35

For these reasons, the best estimates of lower-crustal composition rely on combining seismic velocities of the lower crust with compositions of ‘typical’ lower-crustal lithologies to derive the bulk composition (Christensen and Mooney, 1995; Rudnick and Fountain, 1995; Wedepohl, 1995; Gao et al., 1998a). Wedepohl (1995) used seismic data from the European Geotraverse and Gao et al. (1998a) used data from the North China craton to estimate lower-continental crust composition. The resulting compositions derived from these studies (Table 7, and Figures 12 and 13) reflect the thin and more evolved crust in these regions relative to global averages, and produce too much heat to be representative of global lower crust (Rudnick et al., 1998; see Chapter 4.2). The studies of Rudnick and Fountain (1995) and Christensen and Mooney (1995) are probably best representative of the global deep continental crust, as these authors used overlapping, but not identical, global seismic data sets and independent geochemical data sets to derive the bulk-crust composition. Christensen and Mooney (1995) do not provide the compositional data they used for their lower-crustal assemblages and they do not report a lower-crust composition; thus, one cannot derive an independent estimate of the bulk lower crust from their work. However, they do model the global lower crust (2540 km depth) as containing
7% tonalite gneiss and 93% mafic lithologies (including amphibolite, mafic granulite, and mafic-garnet granulite). Such a maficbulk composition is consistent with the results of Rudnick and Fountain (1995) (Table 7). Thus, we adopt the lowercrust composition of Rudnick and Fountain (1995) as the best available model of global lower-crust composition, with the proviso that our understanding will evolve as more extensive and detailed information becomes available about the seismic velocity structure of the lower crust. This composition has higher iron, magnesium, and calcium, and considerably lower potassium than most other estimates of the lower continental crust (Figure 12), reflecting the overall high P-wave velocities in the lower crust on a worldwide basis and the use of lower crustal xenoliths to derive the average mafic granulite composition. Reliance on xenolith data also accounts for the lower concentrations of highly incompatible trace elements in this composition (e.g., LREEs, rubidium, caesium, barium, thorium, and uranium) compared to most other estimates of the lower crust (Figure 13). For trace elements not considered by Rudnick and Fountain (1995), we adopt the averages of values given in Gao et al. (1998a), Wedepohl (1995), or studies focused specifically on the lower-crustal composition of particular trace elements (e.g., antimony, arsenic, molybdenum (Sims et al., 1990); boron (Leeman et al., 1992); tungsten (Newsom et al., 1996); rhenium; and osmium (Saal et al., 1998)). The resulting lower-crustal composition is given in Table 8. It is interesting to note that the lower-crust composition of Rudnick and Fountain (1995) is quite similar to that of Taylor and McLennan (1985, 1995). However, this similarity is deceptive as Taylor and McLennan’s ‘lower crust’ actually represents the crust below the ‘upper crust’ or between
10 km depth and the Moho. Thus, Taylor and McLennan’s lower crust is equivalent to the combined middle and lower crust given here, and is considerably less evolved than the crustal models adopted here (see Section 4.1.4 for a description of how Taylor and McLennan’s crust composition was derived).

36

Composition of the Continental Crust

3.0

3.0

Transition metals

2.5 2.0

2.0

1.5

1.5

1.0

1.0

0.5

0.5

(a) 4.0

Mn

Sc

V

Cr

Co

High field-strength

2.5

Ni

Cu

Zn

(b)

Ti

Zr

Hf

Nb

Ta

4.0 Weaver and Tarney

3.5

3.5

3.0

3.0

Gao et al.

2.5

2.5

Taylor and McLennan

2.0

2.0

1.5

1.5

1.0

1.0

0.5

0.5

(c)

Li

Rb

Cs

Sr

Ba

Pb

Th

U

(d)

Rare earth

Shaw et al. Wedepohl Median xenolith

La

Ce

Nd

Sm

Eu

Gd

Yb

Lu

Figure 13 Comparison of different models of the trace-element composition of the lower continental crust. All values normalized to the lower-crust composition of Rudnick and Fountain (1995), which is adopted here as the ‘best estimate’ of the global lower crust. Gray-shaded field represents 30% variation from this value. Trace elements are divided into the following groups: (a) transition metals, (b) high-field strength elements, (c) alkali, alkaline earth, and actinides, and (d) REEs.

In summary, our knowledge of lower-crustal composition, like the middle crust, is limited by the ambiguity in deriving chemical compositions from seismic velocities, the lack of high-quality data for a number of trace elements and by the still fragmentary knowledge of the seismic structure of the continental crust. Although the various lower-crustal compositional models in Table 7 show large variations, the true uncertainty in the global model is likely to fall within the seismologically constrained estimates and thus the uncertainty is on the order of 30% for most major elements. Uncertainties in trace-element abundances are generally higher (Figure 13). Whereas concentrations of the transition metals between the different estimates generally fall within
60%, uncertainties in the highly incompatible trace elements (e.g., caesium and thorium; Figure 13) and highly siderophile elements (e.g., PGE) can be as large as an order of magnitude. Despite these rather large uncertainties, there are some conclusions that can be drawn from this analysis. The lower crust has a mafic composition and is strongly depleted in potassium and other highly incompatible elements relative to higher levels of the crust. The lower crust is LREE enriched and probably has a positive europium anomaly (Figure 14). Like the upper and middle crust, it is also characterized by enrichment in lead relative to cerium and praseodymium and depletion in niobium relative to lanthanum. It is also likely to be enriched in strontium relative to neodymium (Figure 14).

4.1.4

Bulk Crust Composition

The earliest estimates of the continental crust composition were derived from analyses and observed proportions of upper

crustal rock types (Clarke, 1889; Clarke and Washington, 1924; Ronov and Yaroshevsky, 1967). These estimates do not take into account the changes in both lithological proportions and metamorphic grade that are now recognized to occur with depth in the crust (see Section 4.1.3) and are thus more appropriately regarded as estimates of upper-crust composition. (It is interesting to note, however, the remarkably good correspondence of these earliest estimates with those of today (cf. Tables 1 and 9)). Taylor (1964) used a different approach to estimate bulk crust composition. Following Goldschmidt (1933), he assumed that the nearly constant REE pattern of sedimentary rocks reflected the REE pattern of the crust as a whole, and recreated that pattern by mixing ‘average’ felsicand mafic-igneous rocks in approximately equal proportions. His composition (Table 9) is also remarkably similar to more modern estimates for a large number of elements, but like the earliest estimates, is more appropriately considered an upper crustal estimate since sediments derive strictly from upper crustal sources. Following the plate-tectonic revolution, Taylor (1967) modified his crust-composition model. Recognizing that the present site of continental growth is at convergent-plate margins, he developed the ‘island arc’ or ‘andesite’ model for crustal growth and hence crust composition. In this model (Taylor, 1967, 1977), the crust is assumed to have a composition equal to average convergent-margin andesite. Taylor and McLennan (1985) discussed the difficulties with this approach. Moreover, it is now recognized that basalts dominate present intra-oceanic arcs (see Chapter 4.21, and references therein). Crust-composition estimates made since the 1970s derive from a variety of approaches. Smithson (1978) was the first to use seismic velocities to determine the lithological makeup of

37

Composition of the Continental Crust

Table 8 Recommended composition of the lower continental crust. Major elements in weight percent. Trace element concentration units the same as in Table 2 Element

SiO2 TiO2 Al2O3 FeOT MnO MgO CaO Na2O K2O P2O5 Li Be B N F S Cl Sc V Cr Co Ni Cu Zn Ga Ge As Se Br Rb Sr Y Zr Nb Mo Ru Pd

Lower crust

Sourcea

53.4 0.82 16.9 8.57 0.10 7.24 9.59 2.65 0.61 0.10 13 1.4 2 34 570 345 250 31 196 215 38 88 26 78 13 1.3 0.2 0.2 0.3 11 348 16 68 5 0.6 0.75 2.8

1 1 1 1 1 1 1 1 1 1 2 2 3 4 2 2 2 1 1 1 1 1 1 1 1 2 5 2 4 1 1 1 1 1 5 6 4

Element

Ag Cd In Sn Sb I Cs Ba La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu Hf Ta W Re Os Ir Pt Au Hg Tl Pb Bi Th U

Lower crust

Sourcea

65 0.1 0.05 1.7 0.1 0.1 0.3 259 8 20 2.4 11 2.8 1.1 3.1 0.48 3.1 0.68 1.9 0.24 1.5 0.25 1.9 0.6 0.6 0.18 0.05 0.05 2.7 1.6 0.014 0.32 4 0.2 1.2 0.2

2 2 4 2 5 4 1 1 1 1 7 1 1 1 1 1 1 1 1 7 1 1 1 1 8 9 9 10 6 11 2 2 1 2 1 1

1000 Lower crust 100

a

Sources: 1. Rudnick and Fountain (1995). 2. Average of values given in Wedepohl (1995) and Gao et al. (1998a). 3. Leeman et al. (1992). 4. Wedepohl (1995). 5. Calculated assuming As/Ce ¼ 0.01, Sb/Ce ¼ 0.005 and Mo/Ce ¼ 0.03 (Sims et al., 1990). 6. Assuming Ru/Ir ratio and Pt/Pd ratios equal to that of upper continental cust. 7. Value interpolated from REE pattern. 8. Average of all values in Table 7, plus correlation from Newsom et al. (1996), using W/Th ¼ 0.5. 9. Saal et al. (1998). 10. Taylor and McLennan (1985). 11. Gao et al. (1998a).

the deep crust. His crust composition is similar to other estimates, save for the very high alkali element contents (4 wt% Na2O and 2.7 wt% K2O), which presumably reflects the choice of granitic rocks used in his calculations. Holland and Lambert (1972), Weaver and Tarney (1984), and Shaw et al. (1986) recognized the importance of granulite-facies rocks in the deep crust and based their crustal models on the composition of rocks from high-grade terranes exposed at the Earth’s surface and previous estimates of upper crustal composition. Taylor and McLennan (1985, 1995), like Taylor’s previous estimates (Taylor, 1967, 1977), derived their crust

10

Chondrite normalized

1 La Ce Pr Nd

Sm Eu Gd Tb Dy Ho Er Tm Yb Lu

1000 Weaver and Tarney Shaw et al. Rudnick and Fountain Wedepohl Gao et al. Taylor and McLennan

100

10 Mantle normalized Rb

1 Cs

Th Ba

K U

La Nb

Pb Ce

Sr Pr

Zr Nd

Sm Hf

Eu

Ti

Ho Y

Yb

Figure 14 REE (upper) and multi-element plot (lower) of the compositions of the lower crust given in Table 7. Compositions derived from individual xenolith suites not shown. Chondrite values from Taylor and McLennan (1985) and primitive mantle values from McDonough and Sun (1995).

composition using an approach based on assumptions about its formation processes. They assumed that 75% of the crust grew during the Archean from bimodal volcanism and the remaining 25% originated from post-Archean accretion of island arcs having an average andesite composition (from Taylor, 1977). To constrain the proportions of mafic- to felsic-Archean volcanics, they used heat-flow data from Archean cratons, which is relatively low and uniform at
40 mW m2 (see Chapter 4.2). Assuming that half the surface heat flow derives from the mantle yielded a mafic to felsic proportion of 2 : 1. This dominantly mafic Archean-crustal component is reflected in the major- and trace-element composition of their crust. Their crust has low SiO2 and K2O, high MgO and CaO, and very high FeO content. It also has the highest transition-metal concentrations and lowest incompatible element concentrations of all the models presented in Table 9. Refinements of the Taylor and McLennan (1985) model are provided by McLennan and Taylor (1996) and McLennan (2001b). The latter is a modification of several trace-element abundances in the upper crust and as such, should not affect their compositional model for the bulk crust, which does not rely on their upper crustal composition. Nevertheless, McLennan (2001b) does provide modified bulk-crust estimates for niobium, rubidium, caesium, and tantalum

1 Taylor (1964)

2 Ronov and Yaroshevsky (1967)

3 Holland and Lambert (1972)

4 Smithson (1978)

60.4 1.0 15.6 7.3 0.12 3.9 5.8 3.2 2.5 0.24 48.7 20 2.8 10 20 625 260 130 22 135 100 25 75 55 70 15 1.5 1.8 0.05 2.5 90 375 33 165 20 1.5

62.2 0.8 15.7 6.3 0.10 3.1 5.7 3.1 2.9

62.8 0.7 15.7 5.5 0.10 3.2 6.0 3.4 2.3 0.20 50.9

63.7 0.7 16.0 5.3 0.10 2.8 4.7 4.0 2.7

70 0.20

47.0

49.0

5 Weaver and Tarney (1984)

63.9 0.6 16.3 5.0 0.08 2.8 4.8 4.2 2.1 0.19 50.5

6 Shaw et al. (1986)

64.5 0.7 15.1 5.7 0.09 3.2 4.8 3.4 2.4 0.14 50.1

7 Christensen and Mooney (1995)

62.4 0.9 14.9 6.9 0.10 3.1 5.8 3.6 2.1 0.20 44.8

8 Rudnick and Fountain (1995)

60.1 0.7 16.1 6.7 0.11 4.5 6.5 3.3 1.9 0.20 54.3 11

9.3

56 35

61 503 14 210 13

13 96 90 26 54 26 71

22 131 119 25 51 24 73 16

76 317 26 203 20

58 325 20 123 8d

9 Wedepohl (1995)

62.8 0.7 15.4 5.7 0.10 3.8 5.6 3.3 2.7 54.3 18 2.4 11 60 525 697 472 16 98 126 24 56 25 65 15 1.4 1.7 0.12 1.0 78 333 24 203 19 1.1 0.1 0.4 70 0.10

10 Gao et al. (1998a)

11 Taylor and McLennan (1985, 1995)

64.2 0.8 14.1 6.8 0.12 3.5 4.9 3.1 2.3 0.18 48.3 17 1.7 18

50.9 13 1.5 10

602 283 179 19 128 92 24 46 38 81 18 1.25 3.1 0.13

30 230 185 29 105 75 80 18 1.6 1.0 0.05

69 285 17.5 175 11 0.65

37c 260 20 100 8d 1.0

1.74 52 0.08

57.1 0.9 15.9 9.1 0.18 5.3 7.4 3.1 1.3

1 80 0.10

12 This studya

60.6 0.72 15.9 6.71 0.10 4.66 6.41 3.07 1.81 0.13 55.3 17 1.9 11 56 553 404 244 21.9 138 135 26.6 59 27 72 16 1.3 2.5 0.13 0.88 49 320 19 132 8 0.8 0.57 1.5 56 0.08

Composition of the Continental Crust

SiO2 TiO2 AI2O3 FeOTb MnO MgO CaO Na2O K 2O P2O5 Mg# Li Be B N F S Cl Sc V Cr Co Ni Cu Zn Ga Ge As Se Br Rb Sr Y Zr Nb Mo Ru Pd Ag Cd

Compositional estimates of the bulk continental crust. Major elements in weight percent. Trace element concentration units the same as in Table 2.

38

Table 9

a

0.1 2.0 0.2 0.5 3.0 425 30 60 8.2 28 6 1.2 5.4 0.9 3.0 1.2 2.8 0.48 3.0 0.50 3 2 1.5

40 0.08 0.45 12.5 0.17 9.6 2.7

See Table 10 for derviation of this estimate. Total Fe as FeO. Mg# ¼ molar 100  Mg/(MgþFetot). c Updated by McLennan (2001b). d Up dated by Barth et al. (2000). Major elements recast to 100% anhydrous. b

707 28 57

764

23 4.1 1.09

20 3.9 1.2

0.53

0.24 1.5 0.23 4.7

15 5.7 1.3

2.6 390 18 42

0.56

5 4

20 9 1.8

2.0 0.33 3.7 0.7d

12.6 5.6 1.4

0.05 2.3 0.3 0.8 3.4 584 30 60 6.7 27 5.3 1.3 4.0 0.65 3.8 0.80 2.1 0.30 2.0 0.35 4.9 1.1 1.0 0.4 0.05 0.05 0.4 2.5 0.040 0.52 14.8 0.085 8.5 1.7

1.5 0.2 2.8 614 31.6 60.0 27.4 4.84 1.27 0.82

2.2 0.35 4.71 0.6 0.7

1.81 1.21 0.009 0.39 15 0.27 7.1 1.2

0.05 2.5 0.2 1.5c 250 16 33 3.9 16 3.5 1.1 3.3 0.60 3.7 0.78 2.2 0.32 2.2 0.30 3.0 0.8c 1.0 0.4 0.05 0.10 3.0 0.36 8.0 0.06 4.2 1.1

0.05 1.7 0.2 0.7 2 456 20 43 4.9 20 3.9 1.1 3.7 0.6 3.6 0.77 2.1 0.28 1.9 0.30 3.7 0.7 1 0.19 0.041 0.037 0.5 1.3 0.03 0.50 11 0.18 5.6 1.3

Composition of the Continental Crust

In Sn Sb I Cs Ba La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu Hf Ta W Re Os Ir Pt Au Hg Tl Pb Bi Th U

39

40

Composition of the Continental Crust

Table 10

Recommended composition of the bulk continental crust

Element

Units

SiO2 TiO2 Al2O3 FeOT MnO MgO CaO Na2O K2O P2O5 Li Be B Na F S Cl Sc V Cr Co Ni Cu Zn Ga Ge As Se Bra Rb Sr Y Zr Nb Mo Rua Pd

wt% ” ” ” ” ” ” ” ” ” mg g1 ” ” ” ” ” ” ” ” ” ” ” ” ” ” ” ” ” ” ” ” ” ” ” ” ng g1 ”

60.6 0.7 15.9 6.7 0.10 4.7 6.4 3.1 1.8 0.1 16 1.9 11 56 553 404 244 21.9 138 135 26.6 59 27 72 16 1.3 2.5 0.13 0.88 49 320 19 132 8 0.8 0.6 1.5

Element

Units

Ag Cd In Sn Sb Ia Cs Ba La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu Hf Ta W Rea Osa Ira Pt Au Hg Tl Pb Bi Th U

ng g1 mg g1 ” ” ” ” ” ” ” ” ” ” ” ” ” ” ” ” ” ” ” ” ” ” ” ng g1 ” ” ” ” mg g1 ” ” ” ” ” ”

56 0.08 0.052 1.7 0.2 0.7 2 456 20 43 4.9 20 3.9 1.1 3.7 0.6 3.6 0.77 2.1 0.28 1.9 0.30 3.7 0.7 1 0.188 0.041 0.037 1.5 1.3 0.03 0.50 11 0.18 5.6 1.3

a

Middle crust is not considered due to lack of data. The total-crust composition is calculated according to the upper, middle and lower-crust compositions obtained in this study and corresponding weighing factors of 0.317, 0.296 and 0.388. The weighing factors are based on the layer thickness of the global continental crust, recalculated from crustal structure and areal proportion of various tectonic units given by Rudnick and Fountain (1995).

(and these are dealt with in the footnotes of Table 9). McLennan and Taylor (1996) revisited the heat-flow constraints on the proportions of mafic and felsic rocks in the Archean crust and revised the proportion of Archean-aged crust to propose a more evolved bulk crust composition. This revised composition is derived from a mixture of 60% Archean crust (which is a 50 : 50 mixture of mafic and felsic endmember lithologies), and 40% average-andesite crust of Taylor (1977). McLennan and Taylor (1996) focused on potassium, thorium, and uranium, and did not provide amended values for other elements, although other incompatible elements will be higher (e.g., rubidium, barium, LREEs) and compatible elements lower in a crust composition so revised. More recently, a number of studies have estimated the bulkcrust composition by deriving lithological proportions for

the deep crust from seismic velocities (as discussed in Section 4.1.3) with upper crustal contributions based on data for surface rocks or previous estimates of the upper crust (Christensen and Mooney, 1995; Rudnick and Fountain, 1995; Wedepohl, 1995; Gao et al., 1998a) (Table 9). Like previous estimates of the crust, all of these show intermediate bulk compositions with very similar major-element contents. The greatest differences in major elements between these recent seismologically based estimates are for MgO, CaO, and K2O, which show
30% variation, with the Rudnick and Fountain (1995) estimate having the highest MgO and CaO, and lowest K2O (Table 9, Figure 15). Most trace elements from these estimates fall within 30% total variation as well (Figure 16); the exceptions are trace elements for which very limited data exist (i.e., sulfur, chlorine, arsenic, tin, mercury, bismuth, and the PGEs). Niobium and tantalum also show >30% total variation, but this is due to the very high niobium and tantalum contents of Wedepohl’s estimate, which reflects his reliance on the old Canadian Shield data, for which niobium content is anomalously high (see discussion in Section 4.1.2.1). These elements were also compromised in the Rudnick and Fountain (1995) crust composition, which relied (indirectly) on the Canadian Shield data for the upper crust by adopting the Taylor and McLennan upper-crust composition (see discussions in Plank and Langmuir (1998) and Barth et al. (2000)). The values for niobium and tantalum in the Rudnick and Fountain (1995) model given in Table 9 have been updated by Barth et al. (2000). These values are similar to those of Gao et al. (1998a) and thus the concentrations of these elements are known to within 30% in the bulk crust. Of all the estimates in Table 9, that of Taylor and McLennan (1985, 1995) stands out as being the most mafic overall (Figures 15 and 16). This mafic composition stems from their model for Archean crust, which constitutes 75% of their crust and is composed of a 2 : 1 mixture of mafic- to felsicigneous rocks. This relatively mafic crust composition was necessitated by their inferred low heat production in Archean crust and the inferred large proportion of the Archean-aged crust. However, such a high proportion of mafic rocks in the Archean crust is at odds with seismic data (summarized in Section 4.1.3), which show that the crust of most Archean cratons is dominated by low velocities, implying the presence of felsic (not mafic) compositions, even in the lower crust. In addition, some of the assumptions used by Taylor and McLennan (1985) regarding heat flow are not very robust, as recognized by McLennan and Taylor (1996) (see also discussion in Rudnick et al., 1998). First, the 20 mWm2 of mantle heat flow they assumed for Archean cratons is probably too high (see Chapter 4.2), thus allowing for more heat production in the crust. Second, granulite-facies felsic rocks are often depleted in heat-producing elements (the Scourian granulites are an extreme example), thereby allowing a greater proportion of felsic rocks in the crust. Third, it is unlikely that 75% of the present continents were formed in the Archean (see Chapter 4.10) and, importantly, the observed low surface heat flow that is the rationale for Taylor and McLennan’s (1985, 1995) dominantly mafic crust composition is restricted to Archean cratons, which constitute only
7% of the present continental crust (Goodwin, 1991). Indeed, heat production of the Taylor and McLennan (1985, 1995) crustal model falls

Composition of the Continental Crust

Normalized to R and G

1.4 1.2

Rudnick and Fountain Wedepohl Gao et al. Chirstensen and Mooney Smithson

1.0 0.8 0.6

(a)

Si

Al

Fe

Mg

Ca

Na

K

Mg

Ca

Na

K

Normalized to R and G

1.4 1.2 1.0 0.8

Holland and Lambert Weaver and Tarney Shaw et al.

0.6 (b)

Taylor and McLennan

Si

Al

Fe

Figure 15 Comparison of different estimates of the major-element composition of the bulk continental crust. All data normalized to the new composition given here (Table 10, ‘R&G’); gray shading depicts 10% variation from this composition. (a) Models based on seismological data (Rudnick and Fountain, 1995; Wedepohl, 1995; Gao et al., 1998a; Christensen and Mooney, 1995 and Smithson, 1978). (b) Models based on both surface exposures (Holland and Lambert, 1972; Weaver and Tarney, 1984; Shaw et al., 1986) and Taylor and McLennan’s (1985, 1995) model-generated crust composition.

outside the range estimated for average continental crust by Jaupart and Mareschal (see Chapter 4.2) (i.e., 0.58 mW m2 versus 0.79–0.99 mW m2). The modifications made to this model by McLennan and Taylor (1996) help to reconcile their model with the above observations, but the proportion of mafic rocks in Archean-aged crust still appears to be high (based on observed seismic velocities) and the total heat production (at 0.70 mW m2) may still be somewhat low.

4.1.4.1

A New Estimate of Crust Composition

In column 12 of Table 9 we present a new estimate of the bulk crust composition. This composition derives from our estimates of upper, middle, and lower crust given in Tables 3,5, and 8, mixed in the proportions derived from the global compilation of Rudnick and Fountain (1995): 31.7% upper, 29.6% middle, and 38.8% lower crust. Our new crustal estimate thus relies heavily on the previously derived lower crust of Rudnick and Fountain (1995), the middle crust of Rudnick and Fountain and Gao et al. (1998a) and the new estimate of upper-crust composition provided here (Table 3). The latter is very similar to the upper crust of Taylor and McLennan (1985) (Figure 8), which was used by Rudnick and Fountain (1995) in calculating their bulk crust composition. The main differences lie in the concentrations of K2O, rubidium, niobium, and

41

tantalum, which are lower in the new estimate of the upper continental crust provided here. Accordingly, this new estimate has many similarities with that of Rudnick and Fountain (1995), but contains lower potassium, rubidium, niobium, and tantalum, and considers a wider range of trace elements than given in that model. The heat production of this new estimate is 0.89 mW m2, which falls in the middle of the range estimated for average-crustal heat production by Jaupart and Mareschal (see Chapter 4.2). Figures 15–17 show how this composition compares to other estimates of crust composition. Figure 15 shows that our new composition has generally higher MgO, CaO, and FeO, and lower Na2O and K2O than most other seismically based models. The differences between our model and that of Wedepohl (1995) and Gao et al. (1998a) likely reflect the regional character of these latter models (western Europe, eastern China), where the crust is thinner and more evolved than the global averages (Chirstensen and Mooney, 1995, and Rudnick and Fountain, 1995). The lower MgO and higher alkali elements in Christensen and Mooney’s model compared to ours must stem from the differences in the chemical databases used to construct these two models, as the lithological proportions of the deep crust are very similar (Section 4.1.3). Rudnick and Fountain (1995) (and hence our current composition) used the compositions of lower-crustal xenoliths to constrain the mafic end-member of the deep crust. These xenoliths have high Mg# and low alkalis (Table 7), and thus may be chemically distinct from mafic rocks exposed on the Earth’s surface (the chemical data used by Christensen and Mooney, 1995). The variations between the different seismological-based crust compositions can be considered representative of the uncertainties that exist in our understanding of the bulk crust composition. Some elemental concentrations (e.g., silicon, aluminum, sodium) are known to within 20% uncertainty. The remaining major-element and many trace-element (transition metals, high-field strength elements, most REE) concentrations are known to within 30% uncertainty. Still some trace element concentrations are yet poorly constrained in the crust, including many of the highly siderophile elements (Figure 16).

4.1.4.2

Intracrustal Differentiation

Table 11 provides the composition of the upper, middle, lower, and bulk crust for comparison purposes and Figure 17 compares their respective REE and extended trace-element patterns. The upper crust has a large negative europium anomaly (Eu/Eu*, Table 11) that is largely complemented by the positive europium anomaly of the lower crust; the middle crust has essentially no europium anomaly. Similar complementary anomalies exist for strontium. These features, in addition to the greater LREE enrichment of the upper crust relative to the lower crust, suggests that the upper crust is largely the product of intracrustal magmatic differentiation in the presence of plagioclase (see Taylor and McLennan, 1985; see Chapter 4.11). That is, the upper crust is dominated by granite that differentiated from the lower crust through partial melting, crystal fractionation and mixing processes. The middle crust has an overall trace-element pattern that is very similar to the upper crust, indicating that it too is dominated by

42

Composition of the Continental Crust

2.0

2.0

2.8 1.5

1.5

1.0

1.0

2.2

0.5

0.5

Transition metals (a)

S

Mn

V

Cr

Co

High field-strength Ni

Cu

Zn

(b)

2.0

2.0

1.5

1.5

1.0

1.0 0.5

0.5

Alkali and alkaline earth (c)

Li

Rb

Cs

Sr

Ba

(d)

Ti

Hf

Zr

Nb

Ta

Mo

W

Rudnick and Fountain Taylor and McLennan Wedepohl Gao et al. Rare earth

La

Ce

Nd

Sm

Eu

Gd

Tb

Yb

Lu

2.0

2.0

Actinides and heavy metals 1.5

1.5

1.0

1.0

0.5

0.5

2.3

Siderophile and chalcophile (e)

Tl

Pb

Bi

Th

U

(f)

Pd

Ag

Cd

In

Sn

Sb

Au

Hg

Figure 16 Comparison of the trace-element composition of bulk continental crust from seismological and model-based approaches. All data normalized to the new composition given here (Table 10, “R&G”). Gray shading depicts  30% variation from Rudnick and Gao composition (this work). (a) Transition metals, (b) high-field strength elements, (c) alkali and alkaline earth metals, and (d) REEs, (e) actinides and heavy metals, and (f) siderophile and chalcophile elements.

the products of intracrustal differentiation. All segments of the crust are characterized by an overall enrichment of the most incompatible elements, as well as high La/Nb and low Ce/Pb ratios. These are characteristics of convergent margin magmas (see Chapter 4.16, and references therein) and thus have implications for the processes responsible for generation of the continental crust as discussed in the next section.

4.1.5

Implications of the Crust Composition

Despite the uncertainties in estimating crust composition discussed in the previous section, there are a number of similarities that all crust-compositional models share and these may be important for understanding the origin of the crust. The crust is characterized by an overall intermediate igneous-rock composition, with relatively high Mg#. It is enriched in incompatible elements (Figure 18), and contains up to 50% of the silicate Earth’s budget of these elements (Rudnick and

Fountain, 1995). It is also well established that the crust is depleted in niobium relative to lanthanum, and has a subchondritic Nb/Ta ratio. These features are not consistent with formation of the crust by single-stage melting of peridotitic mantle, as discussed in Rudnick (1995), Kelemen (1995) and Rudnick et al. (2000) (see also Chapter 4.21). If one assumes that the crust grows ultimately by igneous processes (i.e., magmatic transport of mass from the mantle into the crust), then the disparity between crust composition and the composition of primary mantle melts requires the operation of additional processes to produce the present crust composition. As reviewed in Rudnick (1995) and Kelemen (1995), these additional processes could include (but are not limited to): 1. Recycling of mafic/ultramafic lower crust and upper mantle via density foundering (often referred to as delamination within the geochemical literature). In this process, lithologically stratified continental crust is thickened during an orogenic event, causing the mafic lower crust to transform to eclogite, which has a higher density than the underlying mantle peridotite. Provided

43

Composition of the Continental Crust

Table 11 Comparison of the upper, middle, lower and total continental crust compositions recommended here. See Table 2 for Units Element

Upper crust

Middle crust

Lower crust

Total crust

SiO2 TiO2 Al2O3 FeOT MnO MgO CaO Na2O K2O P2O5 Total Mg# Li Be B N F S Cl Sc V Cr Co Ni Cu Zn Ga Ge As Se Br Rb Sr Y Zr Nb Mo Ru Pd Ag Cd In Sn Sb I Cs Ba La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er

66.6 0.64 15.4 5.04 0.10 2.48 3.59 3.27 2.80 0.15 100.05 46.7 24 2.1 17 83 557 621 370 14.0 97 92 17.3 47 28 67 17.5 1.4 4.8 0.09 1.6 84 320 21 193 12 1.1 0.34 0.52 53 0.09 0.056 2.1 0.4 1.4 4.9 628 31 63 7.1 27 4.7 1.0 4.0 0.7 3.9 0.83 2.3

63.5 0.69 15.0 6.02 0.10 3.59 5.25 3.39 2.30 0.15 100.00 51.5 12 2.3 17

53.4 0.82 16.9 8.57 0.10 7.24 9.59 2.65 0.61 0.10 100.00 60.1 13 1.4 2 34 570 345 250 31 196 215 38 88 26 78 13 1.3 0.2 0.2 0.3 11 348 16 68 5 0.6 0.75 2.8 65 0.10 0.05 1.7 0.10 0.14 0.3 259 8 20 2.4 11 2.8 1.1 3.1 0.48 3.1 0.68 1.9

60.6 0.72 15.9 6.71 0.10 4.66 6.41 3.07 1.81 0.13 100.12 55.3 16 1.9 11 56 553 404 244 21.9 138 135 26.6 59 27 72 16 1.3 2.5 0.13 0.88 49 320 19 132 8 0.8 0.57 1.5 56 0.08 0.052 1.7 0.2 0.71 2 456 20 43 4.9 20 3.9 1.1 3.7 0.6 3.6 0.77 2.1

524 249 182 19 107 76 22 33.5 26 69.5 17.5 1.1 3.1 0.064 65 282 20 149 10 0.60 0.76 48 0.061 1.30 0.28 2.2 532 24 53 5.8 25 4.6 1.4 4.0 0.7 3.8 0.82 2.3

(Continued)

Table 11

(Continued)

Element

Upper crust

Middle crust

Lower crust

Total crust

Tm Yb Lu Hf Ta W Re Os Ir Pt Au Hg Tl Pb Bi Th U Eu/Eu* Heat production (mW m3) Nb/Ta Zr/Hf Th/U K/U La/Yb Rb/Cs K/Rb La/Ta

0.30 2.0 0.31 5.3 0.9 1.9 0.198 0.031 0.022 0.5 1.5 0.05 0.9 17 0.16 10.5 2.7 0.72 1.65

0.32 2.2 0.4 4.4 0.6 0.60

0.85 0.66 0.0079 0.27 15.2 0.17 6.5 1.3 0.96 1.00

0.24 1.5 0.25 1.9 0.6 0.60 0.18 0.05 0.05 2.7 1.6 0.014 0.32 4 0.2 1.2 0.2 1.14 0.19

0.28 1.9 0.30 3.7 0.7 1 0.188 0.041 0.037 1.5 1.3 0.03 0.5 11 0.18 5.6 1.3 0.93 0.89

13.4 36.7 3.8 9475 15.4 20 283 36

16.5 33.9 4.9 15 607 10.7 30 296 42

8.3 35.8 6.0 27 245 5.3 37 462 13

12.4 35.5 4.3 12 367 10.6 24 304 29

the right temperatures and viscosities exist (i.e., hot and goey), the base of the lithosphere will sink into the underlying asthenosphere. Numerical simulations of this process show that it is very likely to occur at the time of arc–continent collision (and in fact, may be impossible to avoid)(Jull and Kelemen, 2001). 2. Production of crust from a mixture of silicic melts derived from subducted oceanic crust, and basaltic melts from peridotite. This process is likely to have been more prevalent in a hotter, Archean Earth (although see Chapter 4.21 for an alternative view) and would have involved extensive silicic melt–peridotite reaction as the slab melts traverse the mantle wedge (Kelemen, 1995). The abundance of Archean-aged granitoids of the so-called ‘TTG’ suite (trondhjemite, tonalite, granodiorite) are often cited as the surface manifestations of these processes (Drummond and Defant, 1990; Martin, 1994; see Chapter 4.11). 3. Weathering of the crust, with preferential recycling of Mg  Ca into the mantle via hydrothermally altered mid-ocean ridge basalt (Albarede, 1998; Anderson, 1982). This hypothesis states that during continental weathering, soluble cations such as Ca2þ, Mg2þ, and Naþ are carried to the oceans while silicon and aluminum remain behind in the continental regolith. Whereas other elements (e.g., sodium) may be returned to the continents via arc magmatism, magnesium may be preferentially sequestered into altered seafloor basalts and returned to the mantle via subduction, producing a net change in the crust composition over time. However, one potential problem

44

Composition of the Continental Crust

Upper crust Middle crust Lower crust

100

1000

Total crust

100

10 10

Chondrite normalized 1 (a)

La Ce Pr Nd

Chondrite normalized

Sm Eu Gd Tb Dy Ho Er Tm Yb Lu

1 La Ce Pr Nd

100

100

10

10

Rb Cs

Th Ba

K U

La Nb

Weaver and Tarney Rudnick and Fountain This Study Wedepohl Gao et al. Taylor and McLennan

Mantle normalized

Mantle normalized 1 (b)

Sm Eu Gd Tb Dy Ho Er Tm Yb Lu

1000

1000

Pb Ce

Sr Pr

Zr Nd

Sm Hf

Eu

Ti

Ho Y

Rb

1 Cs

Yb

Figure 17 Comparison of (a) rare-earth and (b) additional trace-element compositions of the upper, middle, and lower crust recommended here. Chondrite values from Taylor and McLennan (1985), mantle-normalizing values from McDonough and Sun (1995).

with this hypothesis is that examination of altered ocean-floor rocks suggests that magnesium may not be significantly sequestered there (see Chapter 4.16). A fourth possibility, that ultramafic cumulates representing the chemical complement to the andesitic crust are present in the uppermost mantle, is not supported by studies of mantle xenoliths, which show a predominance of restitic peridotite over cumulates (e.g., Wilshire et al., 1988). If such cumulates were originally there, they must have been subsequently removed via a process such as density foundering. All of the above processes require return of mafic to ultramafic lithologies to the convecting mantle. These lithologies are the chemical complement of the present-day andesitic crust. Thus crustal recycling, in various forms, must have been important throughout Earth history. Another implication of the distinctive trace element composition of the continental crust is that the primary setting of crust generation is most likely to be that of a convergent margin. The characteristic depletion of niobium relative to lanthanum seen in the crust (Figure 18) is a ubiquitous feature of convergent margin magmas (see review of Kelemen et al. (Chapter 4.21)) and is virtually absent in intraplate magmas. Simple mixing calculations indicate that the degree of niobium depletion seen in the crust suggests that at least 80% of the crust was generated in a convergent margin (Barth et al., 2000; Plank and Langmuir, 1998).

Th Ba

K U

La Nb

Pb Ce

Sr Pr

Zr Nd

Sm Hf

Eu

Ti

Ho Y

Yb

Figure 18 REE (upper) and multi-element plot (lower) of the compositions of the continental crust given in Table 9. Chondrite values from Taylor and McLennan (1985), mantle-normalizing values from McDonough and Sun (1995).

4.1.6

Earth’s Crust in a Planetary Perspective

The other terrestrial planets show a variety of crustal types, but none that are similar to that of the Earth. Mercury has an ancient, heavily cratered crust with a high albedo (see review of Taylor and Scott (Chapter 1.18)). Its brightness plus the detection of sodium, and more recently the refractory element calcium, in the Mercurian atmosphere (Bida et al., 2000) has led to the speculation that Mercury’s crust may be anorthositic, like the lunar highlands (see Taylor, 1992 and references therein). The MESSENGER mission (http://messenger.jhuapl. edu/), currently planned to rendezvous with Mercury in 2007, should considerably illuminate the nature of the crust on Mercury. In contrast to Mercury’s ancient crust, high-resolution radar mapping of Venus’ cloaked surface has revealed an active planet, both tectonically and volcanically (see review of Fegley (Chapter 1.19) and references therein). Crater densities are relatively constant, suggesting a relatively young surface (
300–500 Ma, Phillips et al., 1992; Schaber et al., 1992; Strom et al., 1994). It has been suggested that this statistically random crater distribution may reflect episodes of mantle overturn followed by periods of quiescence (Schaber et al., 1992; Strom et al., 1994). Most Venusian volcanoes appear to erupt basaltic magmas, but a few are pancake-shaped, which may

Composition of the Continental Crust

signify the eruption of a highly viscous lava such as rhyolite (e.g., Ivanov and Head, 1999). The unimodal topography of Venus is distinct from that of the Earth and there appear to be no equivalents to Earth’s oceanic and continental dichotomy. It is possible that the high elevations on Venus were produced tectonically by compression of basaltic rocks made rigid by the virtual absence of water (Mackwell et al., 1998). Of the terrestrial planets, only Mars has the bimodal topographic distribution seen on the Earth (Smith et al., 1999). In addition, evolved igneous rocks, similar to the andesites found in the continents on Earth, have also been observed on the Martian surface, although their significance and relative abundance is a matter of contention (see review by McSween (Chapter 1.22)). However, the bimodal topography of Mars appears to be an ancient feature (Frey et al., 2002), unlike the Earth’s, which is a product of active plate tectonics. It remains to be seen whether the rocks that compose the high-standing southern highlands of Mars bear any resemblance to those of Earth’s continental crust (McLennan, 2001a; Wanke et al., 2001).

4.1.7

Summary

The crust is the Earth’s major repository of incompatible elements and thus factors prominently into geochemical massbalance calculations for the whole Earth. For this reason, and to understand the processes by which it formed, determining the composition of the continental crust has been a popular pursuit of geochemists from the time the first rocks were analyzed. It has been known for over a century that the continental crust has an average composition approximating to andesite (when cast as an igneous rock type) (Clarke, 1889; Clarke and Washington, 1924). The myriad studies on continental crust composition carried out in the intervening years have refined our picture of the crust’s composition, particularly for trace elements. Based on seismic investigations the crust can be divided into three regions: upper, middle, and lower continental crust. The upper crust is the most accessible region of the solid earth and its composition is estimated from both weighted averages of surface samples and studies of shales and loess. The latter is a particularly powerful means of estimating the average upper crustal concentrations of insoluble to moderately soluble trace elements. Most estimates of the major-element composition of the upper continental crust fall within 20% standard deviation of the mean and thus the composition of this important reservoir appears to be reasonably well known. The concentrations of some trace elements also appear to be known to within 20% (most of the transition metals, rubidium, strontium, yttrium, zirconium, niobium, barium, REE, hafnium, tantalum, lead, thorium, and uranium), whereas others are less-precisely known. In particular, very few estimates have been made of the upper crust’s halogen, sulfur, germanium, arsenic, selenium, indium, and platinum-group element concentration. Lacking the access and widescale natural sampling by sediments afforded the upper crust, the composition of the deep crust must be inferred from more indirect means. Both heat flow and seismic velocities have been employed towards

45

this end. Heat flow provides bounds on the potassium, thorium, and uranium content of the crust, and seismic-wave speeds can be interpreted, with some caveats, in terms of rock types, whose compositions are derived from averages of appropriate deep-crustal lithologies. The middle crust is perhaps the least well characterized of the three crustal regions. This is due to the lack of systematic geochemical studies of amphibolite-facies crustal lithologies. In contrast, the lower crust has been the target of a number of geochemical investigations, yet there is wide variation in different estimates of lower-crust composition. This reflects, in part, the highly heterogeneous character of this part of the Earth. However, some generalities can be made. Heat production must decrease and seismic velocities are observed to increase with depth in the crust. Thus the lower crust is, on an average, mafic in composition and depleted in heat-producing elements. Curiously, the lower crust of many Archean cratons, where heat flow is lowest, has relatively slow P-wave velocities. Such low velocities imply the dominance of evolved rock types and thus these rocks must be highly depleted in potassium, thorium, and uranium compared to their upper crustal counterparts. The andesitic continental crust composition is difficult to explain if the crust is generated by single-stage melting of peridotitic mantle, and additional processes must therefore be involved in its generation. All of these processes entail return of mafic or ultramafic crustal material (which is complementary to the present continental crust) to the convecting mantle. Thus crustal recycling, in various forms, must have been important throughout Earth history and is undoubtedly related to the plate-tectonic cycle on our planet. Crustal recycling, along with the presence of abundant water to facilitate melting (Campbell and Taylor, 1985), may be the major factor responsible for our planet’s unique crustal dichotomy.

Acknowledgments We thank Sandy Romeo and Yongshen Liu for their able assistance in the preparation of this manuscript and updating of the lower-crustal xenolith database. Reviews by Kent Condie and Scott McLennan and comments from Herbert Palme and Rich Walker improved the presentation. This work was supported by NSF grant EAR 99031591 to R.L.R., a National Nature Science Foundation of China grant (40133020) and a Chinese Ministry of Science and Technology grant (G1999043202) to S.G.

References Albare`de F (1998) The growth of continental crust. In: Vauchez A and Meissner RO (eds.) Continents and Their Mantle Roots, Tectonophysics, vol. 296, 1–2, pp. 1–14. Amsterdam: Elsevier. Alirezaei S and Cameron EM (2002) Mass balance during gabbro-amphibolite transition, Bamble sector, Norway: Implications for petrogenesis and tectonic setting of the gabbros. Lithos 60(1–2): 21–45. Anderson AT (1982) Parental basalts in subduction zones: Implications for continental evolution. Journal of Geophysical Research 87: 7047–7060. Aoki K-I (1971) Petrology of mafic inclusion from Itinome-gata, Japan. Contributions to Mineralogy and Petrology 30: 314–331. Arculus RJ and Smith D (1979) Eclogite, pyroxenite, and amphibolite inclusions in the Sullivan Buttes latite, Chino Valley, Yavapai County, Arizona. In: Boyd FR and

46

Composition of the Continental Crust

Meyer HOA (eds.) The Mantle Sample: Inclusions in Kimberlites and Other Volcanics, pp. 309–317. Washington, DC: American Geophysical Union. Arculus RJ, Ferguson J, Chappell BW, et al. (1988) Trace element and isotopic characteristics of eclogites and other xenoliths derived from the lower continental crust of southeastern Australia and southwestern Colorado Plateau USA. In: Smith DC (ed.) Eclogites and Eclogite-Facies Rocks, pp. 335–386. Amsterdam: Elsevier. Baldridge WS (1979) Mafic and ultramafic inclusion suites from the Rio Grande Rift (New Mexico) and their bearing on the composition and thermal state of the lithosphere. Journal of Volcanology and Geothermal Research 6: 319–351. Barth M, McDonough WF, and Rudnick RL (2000) Tracking the budget of Nb and Ta in the continental crust. Chemical Geology 165: 197–213. Barth M, Rudnick RL, Carlson RW, Horn I, and McDonough WF (2002) Re–Os and U–Pb geochronological constraints on the eclogite-tonalite connection in the Archean Man Shield, West Africa. Precambrian Research 118(3–4): 267–283. Beck SL and Zandt G (2002) The nature of orogenic crust in the central Andes. Journal of Geophysical Research 107(B10): 2230. Behn MD and Kelemen PB (2003) The relationship between seismic P-wave velocity and the composition of anhydrous igneous and meta-igneous rocks. Geochemistry, Geophysics, Geosystems (4): 1041. http://dx.doi.org/10.1029/2002GC000393. Berckhemer H (1969) Direct evidence for the composition of the lower crust and Moho. Tectonophysics 8: 97–105. Berg JH, Moscati RJ, and Herz DL (1989) A petrologic geotherm from a continental rift in Antarctica. Earth and Planetary Science Letters 93: 98–108. Bida TA, Killen RM, and Morgan TH (2000) Discovery of calcium in Mercury’s atmosphere. Nature 404: 159–161. Bohlen SR and Mezger K (1989) Origin of granulite terranes and the formation of the lowermost continental crust. Science 244: 326–329. Borodin LS (1998) Estimated chemical composition and petrochemical evolution of the upper continental crust. Geochemistry International 37(8): 723–734. Bowring SA and Williams IS (1999) Priscoan (4.00–4.03 Ga) orthogneisses from northwestern Canada. Contributions to Mineralogy and Petrology 134(1): 3–16. Bradley SD and McCallum ME (1984) Granulite facies and related xenoliths from Colorado-Wyoming kimberlite. In: Kornprobst J (ed.) Kimberlites: II. The Mantle and Crust–Mantle Relationships, vol. 11B, pp. 205–218. Amsterdam: Elsevier. Broadhurst JR (1986) Mineral reactions in xenoliths from the Colorado Plateau: Implications for lower crustal conditions and fluid composition. In: Dawson JB, Carswell DA, Hall J, and Wedepohl KH (eds.) The Nature of the Lower Continental Crust, Geological Society Special Publication 24, pp. 331–349. London, Geological Society of London. Brittan J and Warner M (1996) Seismic velocity, heterogeneity, and the composition of the lower crust. Tectonophysics 264: 249–259. Brittan J and Warner M (1997) Wide-angle seismic velocities in heterogeneous crust. Geophysical Journal International 129: 269–280. Burke MM and Fountain DM (1990) Seismic properties of rocks from an exposure of extended continental crust—new laboratory measurements from the Ivrea zone. Tectonophysics 182: 119–146. Cameron KL and Ward RL (1998) Xenoliths of Grenvillian granulite basement constrain models for the origin of voluminous Tertiary rhyolites, Davis Mountains, west Texas. Geology 26(12): 1087–1090. Cameron KL, Robinson JV, and Niemeyer S, et al. (1992) Contrasting styles of preCenozoic and mid-Tertiary crustal evolution in northern Mexico: Evidence from deep crustal xenoliths from La Olivina. Journal of Geophysical Research 97: 17353–17376. Campbell IH and Taylor SR (1985) No water, no granites—no oceans, no continents. Geophysical Research Letters 10: 1061–1064. Chen S, O’Reilly SY, and Zhou X, et al. (2001) Thermal and petrological structure of the lithosphere beneath Hannuoba, Sino-Korean craton, China: Evidence from xenoliths. Lithos 56: 267–301. Chen W and Arculus RJ (1995) Geochemical and isotopic characteristics of lower crustal xenoliths, San Francisco volcanic field, Arizona, USA. Lithos 36(3–4): 203–225. Chen YD, O’Reilly SY, Kinny PD, and Griffin WL (1994) Dating lower crust and upper-mantle events—an ion microprobe study of xenoliths from kimberlitic pipes, South-Australia. Lithos 32(1–2): 77–94. Chen YD, O’Reilly SY, Griffin WL, and Krogh TE (1998) Combined U–Pb dating and Sm–Nd studies on lower crustal and mantle xenoliths from the Delegate basaltic pipes, southeastern Australia. Contributions to Mineralogy and Petrology 130(2): 154–161. Christensen NI (1982) Vol. II, Seismic velocities. In: Carmichael RS (ed.) Handbook of Physical Properties of Rocks, pp. 1–228. Boca Raton, FL: CRC Press. Christensen NI and Mooney WD (1995) Seismic velocity structure and composition of the continental crust: A global view. Journal of Geophysical Research 100(B7): 9761–9788. Clarke FW (1889) The relative abundance of the chemical elements. Philosophical Society of Washington Bulletin XI: 131–142.

Clarke FW and Washington HS (1924) The composition of the Earth’s crust. US Geological Survey Professional Paper 127. Clitheroe G, Gudmundsson O, and Kennett BLN (2000) The crustal thickness of Australia. Journal of Geophysical Research 105(B6): 13697–13713. Cogley JG (1984) Continental margins and the extent and number of the continents. Reviews of Geophysics and Space Physics 22: 101–122. Cohen RS, O’Nions RK, and Dawson JB (1984) Isotope geochemistry of xenoliths from East Africa: Implications for development of mantle reservoirs and their interaction. Earth and Planetary Science Letters 68: 209–220. Collerson KD, Hearn BC, MacDonald RA, Upton BF, and Park JG (1988) Granulite xenoliths from the Bearpaw Mountains, Montana: Constraints on the character and evolution of lower continental crust. Terra Cognita 8: 270. Condie KC (1993) Chemical composition and evolution of the upper continental crust: Contrasting results form surface samples and shales. Chemical Geology 104: 1–37. Condie KC (1997) Plate Tectonics and Crustal Evolution. Oxford, UK: Butterworth-Heinemann. Condie KC and Selverstone J (1999) The crust of the Colorado plateau: New views of an old arc. Journal of Geology 107(4): 387–397. Condie KC, Latysh N, Van Schmus WR, Kozuch M, and Selverstone J (1999) Geochemistry, Nd and Sr isotopes, and U/Pb zircon ages of Navajo Volcanic Field, Four Corners area, southwestern United States. Chemical Geology 156(1–4): 95–133. Davis GL (1977) The ages and uranium contents of zircons from kimberlites and associated rocks. Carnegie Institution of Washington Yearbook 76: 631–635. Davis GL and Grew ES (1977) Age of zircon from a crustal xenolith, Kilbourne Hole, New Mexico. Carnegie Institution of Washington Yearbook 77: 897–898. Davis WJ (1997) U–Pb zircon and rutile ages from granulite xenoliths in the Slave province: Evidence for mafic magmatism in the lower crust coincident with Proterozoic dike swarms. Geology 25(4): 343–346. Dawson JB (1977) Sub-cratonic crust and upper-mantle models based on xenolith suites in kimberlite and nephelinitic diatremes. Journal of the Geological Society 134: 173–184. Dawson JB and Smith JV (1987) Reduced sapphirine granulite xenoliths from the Lace Kimberlite, South Africa: Implications for the deep structure of the Kaapvaal Craton. Contributions to Mineralogy and Petrology 95: 376–383. Dawson JB, Harley SL, Rudnick RL, and Ireland TR (1997) Equilibration and reaction in Archaean quartz-sapphirine granulite xenoliths from the Lace Kimberlite pipe, South Africa. Journal of Metamorphic Geology 15(2): 253–266. Dessai AG and Vaselli O (1999) Petrology and geochemistry of xenoliths in lamprophyres from the Deccan Traps: Implications for the nature of the deep crust boundary in western India. Mineralogical Magazine 63(5): 703–722. Dessai AG, Knight K, and Vaselli O (1999) Thermal structure of the lithosphere beneath the Deccan Trap along the western Indian continental margin: Evidence from xenolith data. Journal of the Geological Society of India 54(6): 585–598. Dobosi G, Kempton PD, Downes H, Embey-Isztin A, Thirlwall M, and Greenwood P (2003) Lower crustal granulite xenoliths from the Pannonian Basin, Hungary: Part 2. Sr–Nd–Pb–Hf and O isotope evidence for formation of continental lower crust by tectonic emplacement of oceanic crust. Contributions to Mineralogy and Petrology 144: 671–683. Dodge FCW, Lockwood JP, and Calk LC (1988) Fragments of the mantle and crust from beneath the Sierra Nevada batholith: Xenoliths in a volcanic pipe near Big Creek, California. Geological Society of America Bulletin 100: 938–947. Dodge FW, Calk LC, and Kistler RW (1986) Lower crustal xenoliths, Chinese Peak lava flow, central Sierra Nevada. Journal of Petrology 27: 1277–1304. Domenick MA, Kistler RW, Dodge FW, and Tatsumoto M (1983) Nd and Sr isotopic study of crustal and mantle inclusions from the Sierra Nevada and implications for batholith petrogenesis. Geological Society of America Bulletin 94: 713–719. Dostal J and Capedri S (1979) Rare earth elements in high-grade metamorphic rocks from the western Alps. Lithos 12: 41–49. Dostal J, Dupuy C, and Leyreloup A (1980) Geochemistry and petrology of meta-igneous granulitic xenoliths in Neogene volcanic rocks of the Massif Central, France—implications for the lower crust. Earth and Planetary Science Letters 50: 31–40. Downes H (1993) The nature of the lower continental crust of Europe–petrological and geochemical evidence from xenoliths. Physics of the Earth and Planetary Interiors 79(1–2): 195–218. Downes H and Leyreloup AF (1986) Granulitic xenoliths from the French Massif Central—petrology, Sr and Nd isotope systematics and model age estimates. In: Dawson B, Carswell DA, Hall J, and Wedepohl KH (eds.) The Nature of the Lower Continental Crust, pp. 319–330. London: Geological Society of London. Downes H, Dupuy C, and Leyreloup AF (1990) Crustal evolution of the Hercynian belt of western Europe: Evidence from lower-crustal granulitic xenoliths (French Massif-Central). Chemical Geology 83(3–4): 209–231.

Composition of the Continental Crust

Downes H, Kempton PD, Briot D, Harmon RS, and Leyreloup AF (1991) Pb and O isotope systematics in granulite facies xenoliths, French Massif-Central—implications for crustal processes. Earth and Planetary Science Letters 102(3–4): 342–357. Downes H, Peltonen P, Manttari I, and Sharkov EV (2002) Proterozoic zircon ages from lower crustal granulite xenoliths, Kola Peninsula, Russia: Evidence for crustal growth and reworking. Journal of the Geological Society 159: 485–488. Drummond BJ (1988) A review of crust upper mantle structure in the Precambrian areas of Australia and implications for Precambrian crustal evolution. Precambrian Research 40(1): 101–116. Drummond MS and Defant MJ (1990) A model for trondhjemite-tonalite-dacite genesis and crustal growth via slab melting: Archean to modern comparisons. Journal of Geophysical Research 95(B13): 21503–21521. Ducea M (2001) The California arc: Thick granitic batholiths, eclogitic residues, lithospheric-scale thrusting, and magmatic flare-ups. GSA Today 11(11): 4–10. Ducea MN and Saleeby JB (1996) Buoyancy sources for a large, unrooted mountain range, the Sierra Nevada, California: Evidence from xenolith thermobarometry. Journal of Geophysical Research 101(B4): 8229–8244. Ducea MN and Saleeby JB (1998) The age and origin of a thick mafic-ultramafic keel from beneath the Sierra Nevada batholith. Contributions to Mineralogy and Petrology 133(1–2): 169–185. Durrheim RJ and Green RWE (1992) A seismic refraction investigation of the Archaean Kaapvaal craton, South Africa, using mine tremors as the energy source. Geophysical Journal International 108: 812–832. Durrheim RJ and Mooney WD (1994) The evolution of the Precambrian lithosphere: Seismological and geochemical constraints. Journal of Geophysical Research 99: 15359–15374. Eade KE and Fahrig WF (1971) Chemical Evolutionary Trends of Continental Plates—Preliminary Study of the Canadian Shield. Bulletin - Geological Survey of Canada 179, 51p. Eade K E and Fahrig W F (1973) Regional, Lithological, and Temporal Variation in the Abundances of some Trace Elements in the Canadian Shield. Geological Survey of Canada Paper 72–46, Ottawa, Ontario. Eberz GW, Clarke DB, Chatterjee AK, and Giles PS (1991) Chemical and isotopic composition of the lower crust beneath the Meguma Lithotectonic Zone, Nova Scotia: Evidence from granulite facies xenoliths. Contributions to Mineralogy and Petrology 109: 69–88. Edwards AC, Lovering JF, and Ferguson J (1979) High pressure basic inclusions from the Kayrunnera kimberlitic breccia pipe in New South Wales, Australia. Contributions to Mineralogy and Petrology 69: 185–192. Ehrenberg SN and Griffin WL (1979) Garnet granulite and associated xenoliths in minette and serpentinite diatremes of the Colorado Plateau. Geology 7: 483–487. Embey-Isztin A, Scharbert HG, Deitrich H, and Poultidis H (1990) Mafic granulite and clinopyroxenite xenoliths from the Transdanubian volcanic region (Hungary): Implication for the deep structure of the Pannonian Basin. Mineralogical Magazine 54: 463–483. Embey-Isztin A, Downes H, Kempton PD, Dobosi G, and Thirlwall M (2003) Lower crustal granulite xenoliths from the Pannonian Basin, Hungary: Part 1. Mineral chemistry, thermobarometry and petrology. Contributions to Mineralogy and Petrology 144: 652–670. Ertan IE and Leeman WP (1999) Fluid inclusions in mantle and lower crustal xenoliths from the Simcoe volcanic field, Washington. Chemical Geology 154(1–4): 83–95. Esperanc¸a S and Garfunkel Z (1986) Ultramafic xenoliths from the Mt. Carmel area (Karem Maharal Volcano), Israel. Lithos 19: 43–49. Esperanc¸a S, Carlson RW, and Shirey SB (1988) Lower crustal evolution under central Arizona: Sr, Nd, and Pb isotopic and geochemical evidence from the mafic xenoliths of Camp Creek. Earth and Planetary Science Letters 90: 26–40. Esser BK and Turekian KK (1993) The osmium isotopic composition of the continental crust. Geochimica et Cosmochimica Acta 57: 3093–3104. Fahrig WF and Eade KE (1968) The chemical evolution of the Canadian Shield. Geochimica et Cosmochimica Acta 5: 1247–1252. Fountain DM (1976) The Ivrea-Verbano and Strona-Ceneri zones, northern Italy: A cross-section of the continental crust—new evidence from seismic velocities of rock samples. Tectonophysics 33: 145–165. Fountain DM and Salisbury MH (1981) Exposed cross sections through the continental crust: Implications for crustal structure, petrology, and evolution. Earth and Planetary Science Letters 56: 263–277. Fountain DM and Salisbury MH (1995) Seismic properties of rock samples from the Pikwitonei granulite belt—God’s Lake domain crustal cross section, Manitoba. Canadian Journal of Earth Sciences 33(5): 757–768. Fountain DM, Percival J, and Salisbury MH (1990a) Exposed cross sections of the continental crust-synopsis. In: Salisbury MH and Fountain DM (eds.) Exposed Cross-Sections of the Continental Crust, pp. 653–662. Amsterdam: Kluwer.

47

Fountain DM, Salisbury MH, and Percival J (1990b) Seismic structure of the continental crust based on rock velocity measurements from the Kapuskasing uplift. Journal of Geophysical Research 95: 1167–1186. Fountain DM, Salisbury MH, and Furlong KP (1987) Heat production and thermal conductivity of rocks from the Pikwitonei-Sachigo continental cross section, central Manitoba: Implications for the thermal structure of Archean crust. Canadian Journal of Earth Sciences 24: 1583–1594. Francis DM (1976) Corona-bearing pyroxene granulite xenoliths and the lower crust beneath Nunivak Island, Alaska. Canadian Mineralogist 14: 291–298. Frey HV, Roark JH, Shockey KM, Frey EL, and Sakimoto SEH (2002) Ancient lowlands on Mars. Geophysical Research Letters 29(10) 22-1–22-4. Gallet S, Jahn B-M, van Vliet Lanoe¨ B, Dia A, and Rossello E (1998) Loess geochemistry and its implications for particle origin and composition of the upper continental crust. Earth and Planetary Science Letters 156: 157–172. Gao S, Luo T-C, and Zhang B-R, et al. (1998a) Chemical composition of the continental crust as revealed by studies in east China. Geochimica et Cosmochimica Acta 62: 1959–1975. Gao S, Zhang B-R, Jin Z-M, Kern H, Luo T-C, and Zhao Z-D (1998b) How mafic is the lower continental crust? Earth and Planetary Science Letters 106: 101–117. Gao S, Kern H, and Liu YS, et al. (2000) Measured and calculated seismic velocities and densities for granulites from xenolith occurrences and adjacent exposed lower crustal sections: A comparative study from the North China craton. Journal of Geophysical Research 105(B8): 18965–18976. Goldschmidt VM (1933) Grundlagen der quantitativen Geochemie. Fortschritte der Mineralogie, Kristallographie und Petrographie 17: 112. Goldschmidt VM (1958) Geochemistry. Oxford: Oxford University Press. Goodwin AM (1991) Precambrian Geology. London: Academic Press. Gorman AR, Clowes RM, and Ellis RM, et al. (2002) Deep probe: Imaging the roots of western North America. Canadian Journal of Earth Sciences 39(3): 375–398. Grapes RH (1986) Melting and thermal reconstitution of pelitic xenoliths, Wehr volcano, East Eifel, West Germany. Journal of Petrology 27: 343–396. Gregoire M, Mattielli N, and Nicollet C, et al. (1994) Oceanic mafic granulite xenoliths from the Kerguelen archipelago. Nature 367: 360–363. Gregoire M, Cottin JY, Giret A, Mattielli N, and Weis D (1998) The meta-igneous granulite xenoliths from Kerguelen archipelago: Evidence of a continent nucleation in an oceanic setting. Contributions to Mineralogy and Petrology 133(3): 259–283. Griffin WL and O’Reilly SY (1986) The lower crust in eastern Australia: Xenolith evidence. In: Dawson B, Carswell DA, Hall J, and Wedepohl KH (eds.) The Nature of the Lower Continental Crust, Geological Society Special Publication 25, pp. 363–374. London, Geological Society of London. Griffin WL and O’Reilly SY (1987) Is the continental Moho the crust–mantle boundary? Geology 15: 241–244. Griffin WL, Carswell DA, and Nixon PH (1979) Lower-crustal granulites and eclogites from Lesotho, southern Africa. In: Boyd FR and Meyer HOA (eds.) The Mantle Sample: Inclusions in Kimberlites, pp. 59–86. Washington, DC: American Geophysical Union. Griffin WL, Sutherland FL, and Hollis JD (1987) Geothermal profile and crust–mantle transition beneath east-central Queensland: Volcanology, xenolith petrology and seismic data. Journal of Volcanology and Geothermal Research 31: 177–203. Griffin WL, Jaques AL, Sie SH, Ryan CG, Cousens DR, and Suter GF (1988) Conditions of diamond growth: A proton microprobe study of inclusions in West Australian diamonds. Contributions to Mineralogy and Petrology 99: 143–158. Hacker BR, Gnos E, Ratschbacher L, et al. (2000) Hot and dry deep crustal xenoliths from Tibet. Science 287(5462): 2463–2466. Halliday AN, Aftalion M, Upton BGJ, Aspen P, and Jocelyn J (1984) U–Pb isotopic ages from a granulite-facies xenolith from Partan Craig in the Midland Valley of Scotland. Transactions of the Royal Society of Edinburgh: Earth and Environmental Science 75: 71–74. Hanchar JM, Miller CF, Wooden JL, Bennett VC, and Staude J-MG (1994) Evidence from xenoliths for a dynamic lower crust, eastern Mojave desert, California. Journal of Petrology 35: 1377–1415. Harley SL (1989) The origin of granulites: A metamorphic perspective. Geological Magazine 126: 215–247. Hart RJ, Nicolaysen LO, and Gale NH (1981) Radioelement concentrations in the deep profile through Precambrian basement of the Vredefort structure. Journal of Geophysical Research 86: 10639–10652. Hart RJ, Andreoli MAG, Tredoux M, and Dewit MJ (1990) Geochemistry across an exposed section of Archean crust at Vredefort, South Africa with implications for midcrustal discontinuities. Chemical Geology 82(1–2): 21–50. Haskin MA and Haskin LA (1966) Rare earths in European shales: A redetermination. Science 154: 507–509. HaskinL A, Wildeman TR, Frey FA, Collins KA, Keedy CR, and Haskin MA (1966) Rare earths in sediments. Journal of Geophysical Research 71(24): 6091–6105.

48

Composition of the Continental Crust

Hattori Y, Suzuki K, Honda M, and Shimizu H (2003) Re–Os isotope systematics of the Taklimakan Desert sands, moraines and river sediments around the Taklimakan Desert, and of Tibetan soils. Geochimica et Cosmochimica Acta 67: 1195–1206. Hayob JL, Essene EJ, Ruiz J, Ortega-Gutierrez F, and Aranda-Gomez JJ (1989) Young high-temperature granulites form the base of the crust in central Mexico. Nature 342: 265–268. Heier KS (1973) Geochemistry of granulite facies rocks and problems of their origin. Philosophical Transactions of the Royal Society Series A 273: 429–442. Heinrichs H, Schulz-Dobrick B, and Wedepohl KH (1980) Terrestrial geochemistry of Cd, Bi, Tl, Pb, Zn, and Rb. Geochimica et Cosmochimica Acta 44: 1519–1533. Hickey-Vargas R, Abdollahi MJ, Parada MA, Lopezescobar L, and Frey FA (1995) Crustal xenoliths from Calbuco volcano, Andean southern volcanic zone—implications for crustal composition and magma-crust interaction. Contributions to Mineralogy and Petrology 119(4): 331–344. Holbrook WS and Kelemen PB (1993) Large igneous province on the US Atlantic margin and implications for magmatism during continental breakup. Nature 364: 433–436. Holbrook WS, Mooney WD, and Christensen NI (1992) The seismic velocity structure of the deep continental crust. In: Fountain DM, Arculus R, and Kay RW (eds.) Continental Lower Crust, pp. 1–44. Amsterdam: Elsevier. Holland JG and Lambert RSJ (1972) Major element chemical composition of shields and the continental crust. Geochimica et Cosmochimica Acta 36: 673–683. Ho¨ltta¨ P, Huhma H, Manttari I, Peltonen P, and Juhanoja J (2000) Petrology and geochemistry of mafic granulite xenoliths from the Lahtojoki kimberlite pipe, eastern Finland. Lithos 51(1–2): 109–133. Hu S, He L, and Wang J (2000) Heat flow in the continental area of China: A new data set. Earth and Planetary Science Letters 179: 407–419. Huang YM, van Calsteren P, and Hawkesworth CJ (1995) The evolution of the lithosphere in southern Africa—a perspective on the basic granulite xenoliths from kimberlites in South-Africa. Geochimica et Cosmochimica Acta 59(23): 4905–4920. Hunter RH, Upton BGJ, and Aspen P (1984) Meta-igneous granulite and ultramafic xenoliths from basalts of the Midland Valley of Scotland: Petrology and mineralogy of the lower crust and upper mantle. Transactions of the Royal Society of Edinburgh 75: 75–84. Ivanov MA and Head JW (1999) Stratigraphic and geographic distribution of steep-sided domes on Venus: Preliminary results from regional geological mapping and implications for their origin. Journal of Geophysical Research 104(E8): 18907–18924. Jahn BM, Gallet S, and Han JM (2001) Geochemistry of the Xining, Xifeng, and Jixian sections, Loess Plateau of China: Eolian dust provenance and paleosol evolution during the last 140 ka. Chemical Geology 178(1–4): 71–94. James DE, Padovani ER, and Hart SR (1980) Preliminary results on the oxygen isotopic composition of the lower crust, Kilbourne Hole Maar, New Mexico. Geophysical Research Letters 7: 321–324. Jones AP, Smith JV, Dawson JB, and Hansen EC (1983) Metamorphism, partial melting, and K-metasomatism of garnet-scapolite-kyanite granulite xenoliths from Lashaine, Tanzania. Journal of Geology 91: 143–166. Jull M and Kelemen PB (2001) On the conditions for lower crustal convective instability. Journal of Geophysical Research 106(4): 6423–6446. Kalamarides RI, Berg JH, and Hank RA (1987) Lateral isotopic discontinuity in the lower crust: An example from Antarctica. Science 237: 1192–1195. Kay RW and Kay SM (1981) The nature of the lower continental crust: Inferences from geophysics, surface geology, and crustal xenoliths. Reviews of Geophysics and Space Physics 19: 271–297. Kay RW and Kay SM (1991) Creation and destruction of lower continental crust. Geologische Rundschau 80: 259–278. Kay SM and Kay RW (1983) Thermal history of the deep crust inferred from granulite xenoliths, Queensland, Australia. American Journal of Science 283: 486–513. Kelemen PB (1995) Genesis of high Mg# andesites and the continental crust. Contributions to Mineralogy and Petrology 120: 1–19. Kempton PD and Harmon RS (1992) Oxygen-isotope evidence for large-scale hybridization of the lower crust during magmatic underplating. Geochimica et Cosmochimica Acta 55: 971–986. Kempton PD, Harmon RS, Hawkesworth CJ, and Moorbath S (1990) Petrology and geochemistry of lower crustal granulites from the Geronimo volcanic field, southeastern Arizona. Geochimica et Cosmochimica Acta 54: 3401–3426. Kempton PD, Downes H, and Sharkov EV, et al. (1995) Petrology and geochemistry of xenoliths from the northern Baltic shield: Evidence for partial melting and metasomatism in the lower crust beneath an Archaean terrane. Lithos 36(3–4): 157–184. Kempton PD, Downes H, and Embey-Isztin A (1997) Mafic granulite xenoliths in Neogene alkali basalts from the western Pannonian Basin: Insights into the lower crust of a collapsed orogen. Journal of Petrology 38(7): 941–970.

Kempton PD, Downes H, Neymark LA, Wartho JA, Zartman RE, and Sharkov EV (2001) Garnet granulite xenoliths from the northern Baltic shield the underplated lower crust of a palaeoproterozoic large igneous province. Journal of Petrology 42(4): 731–763. Kern H, Gao S, and Liu Q-S (1996) Seismic properties and densities of middle and lower crustal rocks exposed along the North China geoscience transect. Earth and Planetary Science Letters 139: 439–455. Ketcham RA (1996) Distribution of heat-producing elements in the upper and middle crust of southern and west central Arizona: Evidence from core complexes. Journal of Geophysical Research 101(B6): 13611–13632. Kopylova MG, O’Reilly SY, and Genshaft YS (1995) Thermal state of the lithosphere beneath Central Mongolia: Evidence from deep-seated xenoliths from the Shavaryn-Saram volcanic centre in the Tariat depression, Hangai, Mongolia. Lithos 36: 243–255. Kuno H (1967) Mafic and ultramafic nodules from Itinome-gata, Japan. In: Wiley PJ (ed.) Ultramafic and Related Rocks, pp. 337–342. New York: Wiley. Kyle PR, Wright A, and Kirsch I (1987) Ultramafic xenoliths in the late Cenozoic McMurdo volcanic group, western Ross Sea embayment, Antarctica. In: Nixon PH (ed.) Mantle Xenoliths, pp. 287–294. New York: Wiley. Le Bas MJ and Streckeisen AL (1991) The IUGS systematics of igneous rocks. Journal of the Geological Society 148: 825–833. Lee C-Y, Chung SL, Chen C-H, and Hsieh YL (1993) Mafic granulite xenoliths from Penghu Islands: Evidence for basic lower crust in SE China continental margin. Journal of the Geological Society of China 36(4): 351–379. Leech ML (2001) Arrested orogenic development: Eclogitization, delamination, and tectonic collapse. Earth and Planetary Science Letters 185(1–2): 149–159. Leeman WP, Menzies MA, Matty DJ, and Embree GF (1985) Strontium, neodymium, and lead isotopic compositions of deep crustal xenoliths from the Snake River Plain: Evidence for Archean basement. Earth and Planetary Science Letters 75: 354–368. Leeman WP, Sisson VB, and Reid MR (1992) Boron geochemistry of the lower crust-evidence from granulite terranes and deep crustal xenoliths. Geochimica et Cosmochimica Acta 56(2): 775–788. LePichon X, Henry P, and Goffe B (1997) Uplift of Tibet: From eclogites to granulitesimplications for the Andean Plateau and the Variscan belt. Tectonophysics 273(1–2): 57–76. Leyreloup A, Dupuy C, and Andriambololona R (1977) Catazonal xenoliths in French Neogene volcanic rocks: Constitution of the lower crust: 2. Chemical composition and consequences of the evolution of the French Massif Central Precambrian crust. Contributions to Mineralogy and Petrology 62: 283–300. Leyreloup A, Bodinier JL, Dupuy C, and Dostal J (1982) Petrology and geochemistry of granulite xenoliths from Central Hoggar (Algeria)—implications for the lower crust. Contributions to Mineralogy and Petrology 79: 68–75. Liu YS, Gao S, Jin SY, et al. (2001) Geochemistry of lower crustal xenoliths from Neogene Hannuoba basalt, North China craton: Implications for petrogenesis and lower crustal composition. Geochimica et Cosmochimica Acta 65(15): 2589–2604. Lombardo B and Rolfo F (2000) Two contrasting eclogite types in the Himalayas: Implications for the Himalayan orogeny. Journal of Geodynamics 30(1–2): 37–60. Loock G, Seck HA, and Stosch H-G (1990) Granulite facies lower crustal xenoliths from the Eifel, West Germany: Petrological and geochemical aspects. Contributions to Mineralogy and Petrology 105: 25–41. Lovering JF and White AJR (1964) The significance of primary scapolite in granulitic inclusions from deep-seated pipes. Journal of Petrology 5: 195–218. Lovering JF and White AJR (1969) Granulitic and eclogitic inclusions from basic pipes at Delegate, Australia. Contributions to Mineralogy and Petrology 21: 9–52. Lucassen F, Lewerenz S, Franz G, Viramonte J, and Mezger K (1999) Metamorphism, isotopic ages and composition of lower crustal granulite xenoliths from the Cretaceous Salta Rift, Argentina. Contributions to Mineralogy and Petrology 134(4): 325–341. Luosto U and Korhonen H (1986) Crustal structure of the baltic shield based on off-fennolora refraction data. Tectonophysics 128: 183–208. Luosto U, Flu¨h ER, Lund C-E, and Group W (1989) The crustal structure along the POLAR profile from seismic refraction investigations. Tectonophysics 162: 51–85. Luosto U, Tiira T, Korhonen H, et al. (1990) Crust and upper mantle structure along the DSS Baltic profile in SE Finland. Geophysical Journal International 101: 89–110. Mackwell SJ, Zimmerman ME, and Kohlstedt DL (1998) High-temperature deformation of dry diabase with application to tectonics on Venus. Journal of Geophysical Research 103(B1): 975–984. Markwick AJW and Downes H (2000) Lower crustal granulite xenoliths from the Arkhangelsk kimberlite pipes: Petrological, geochemical and geophysical results. Lithos 51(1–2): 135–151.

Composition of the Continental Crust

Markwick AJW, Downes H, and Veretennikov N (2001) The lower crust of SE Belarus: Petrological, geophysical, and geochemical constraints from xenoliths. Tectonophysics 339(1–2): 215–237. Martin H (1994) The Archean grey gneisses and the genesis of continental crust. In: Condie KC (ed.) Archean Crustal Evolution, pp. 205–259. Amsterdam: Elsevier. Mattie PD, Condie KC, Selverstone J, and Kyle PR (1997) Origin of the continental crust in the Colorado Plateau: Geochemical evidence from mafic xenoliths from the Navajo volcanic field, southwestern USA. Geochimica et Cosmochimica Acta 61(10): 2007–2021. Mayer A, Mezger K, and Sinigoi S (2000) New Sm–Nd ages for the Ivrea-Verbano Zone, Sesia and Sessera valleys (Northern-Italy). Journal of Geodynamics 30(1–2): 147–166. McBirney AR and Aoki K-I (1973) Factors governing the stability of plagioclase at high pressures as shown by spinel-gabbro xenoliths from the Kerguelen archipelago. American Mineralogist 58: 271–276. McCulloch MT, Arculus RJ, Chappell BW, and Ferguson J (1982) Isotopic and geochemical studies of nodules in kimberlite have implications for the lower continental crust. Nature 300: 166–169. McDonough WF and Sun S-S (1995) Composition of the Earth. Chemical Geology 120: 223–253. McDonough WF, Rudnick RL, and McCulloch MT (1991) The chemical and isotopic composition of the lower eastern Australian lithosphere: A review. In: Drummond B (ed.) The Nature of the Eastern Australian Lithosphere, Geological Society of Australia Special Publication 17, pp.163–188. Sydney. McDonough WF, Sun S-S, Ringwood AE, Jagoutz E, and Hofmann AW (1992) Potassium, rubidium, and cesium in the Earth and Moon and the evolution of the mantle of the Earth. Geochimica et Cosmochimica Acta 56: 1001–1012. McLennan SM (2001a) Crustal heat production and the thermal evolution of Mars. Geophysical Research Letters 28(21): 4019–4022. McLennan SM (2001b) Relationships between the trace element composition of sedimentary rocks and upper continental crust. Geochemistry, Geophysics, Geosystems 2: 1021. McLennan SM and Taylor SR (1996) Heat flow and the chemical composition of continental crust. Journal of Geology 104: 377–396. McLennan SM, Nance WB, and Taylor SR (1980) Rare earth element-thorium correlations in sedimentary rocks, and the composition of the continental crust. Geochimica et Cosmochimica Acta 44: 1833–1839. Mehnert KR (1975) The Ivrea zone: A model of the deep crust. Neues Jahrbuch fu¨r Mineralogie (Abhandlungen) 125: 156–199. Meltzer A and Christensen N (2001) Nanga Parbat crustal anisotrophy: Implication for interpretation of crustal velocity structure and shear-wave splitting. Geophysical Research Letters 28(10): 2129–2132. Mengel K (1990) Crustal xenoliths from Tertiary volcanics of the northern Hessian depression: Petrological and chemical evolution. Contributions to Mineralogy and Petrology 104: 8–26. Mengel K and Wedepohl KH (1983) Crustal xenoliths in Tertiary volcanics from the northern Hessian depression. In: Fuchs K, et al. (ed.) Plateau Uplift, pp. 332–335. Berlin: Springer. Mengel K, Sachs PM, Stosch HG, Worner G, and Loock G (1991) Crustal xenoliths from Cenozoic volcanic fields of West Germany implications for structure and composition of the continental crust. Tectonophysics 195(2–4): 271. Meyer HOA and Brookins DG (1976) Sapphirine, sillimanite, and garnet in granulite xenoliths from Stockdale kimberlite, Kansas. American Mineralogist 61: 1194–1202. Miller JD and Christensen NI (1994) Seismic signature and geochemistry of an island arc: A multidisciplinary study of the Kohistan accreted terrane, northern Pakistan. Journal of Geophysical Research 99(B6): 11623–11642. Mittlefehldt DW (1984) Genesis of clinopyroxene-amphibole xenoliths from Birket Ram: Trace element and petrologic constraints. Contributions to Mineralogy and Petrology 88: 280–287. Mittlefehldt DW (1986) Petrology of high pressure clinopyroxenite series xenoliths, Mount Carmel, Israel. Contributions to Mineralogy and Petrology 94: 245–252. Moecher DP, Valley JW, and Essene EJ (1994) Extraction and carbon isotope analysis of CO2 from scapolite in deep crustal granulites and xenoliths. Geochimica et Cosmochimica Acta 58(2): 959–967. Mooney WD and Meissner R (1992) Multi-genetic origin of crustal reflectivity: A review of seismic reflection profiling of the continental lower crust and Moho. In: Fountain DM, Arculus R, and Kay RW (eds.) Continental Lower Crust, pp. 45–80. Elsevier. Moser DE and Heaman LM (1997) Proterozoic zircon growth in Archean lower crust xenoliths, southern Superior craton: A consequence of Matachewan ocean opening. Contributions to Mineralogy and Petrology 128: 164–175. Moser DE, Flowers RM, and Hart RJ (2001) Birth of the Kaapvaal tectosphere 3.08 billion years ago. Science 291(5503): 465–468.

49

Nasir S (1992) The lithosphere beneath the northwestern part of the Arabian plate (Jordan)—evidence from xenoliths and geophysics. Tectonophysics 201(3–4): 357–370. Nasir S (1995) Mafic lower crustal xenoliths from the northwestern part of the Arabian plate. European Journal of Mineralogy 7(1): 217–230. Nasir S and Safarjalani A (2000) Lithospheric petrology beneath the northern part of the Arabian plate in Syria: Evidence from xenoliths in alkali basalts. Journal of African Earth Sciences 30(1): 149–168. Nesbitt HW and Young GM (1984) Prediction of some weathering trends of plutonic and volcanic rocks based on thermodynamic and kinetic considerations. Geochimica et Cosmochimica Acta 48: 1523–1534. Newsom HE, Sims KWW, Noll PD Jr., Jaeger WL, Maehr SA, and Beserra TB (1996) The depletion of tungsten in the bulk silicate Earth: Constraints on core formation. Geochimica et Cosmochimica Acta 60: 1155–1169. Nguuri TK, Gore J, and James DE, et al. (2001) Crustal structure beneath southern Africa and its implications for the formation and evolution of the Kaapvaal and Zimbabwe cratons. Geophysical Research Letters 28(13): 2501–2504. Nimz GJ, Cameron KL, Cameron M, and Morris SL (1986) Petrology of the lower crust and upper mantle beneath southeastern Chihuahua, Mexico. Geof´ı sica Internacional 25: 85–116. Niu FL and James DE (2002) Fine structure of the lowermost crust beneath the Kaapvaal craton and its implications for crustal formation and evolution. Earth and Planetary Science Letters 200(1–2): 121–130. Nyblade AA and Pollack HN (1993) A global analysis of heat flow from Precambrian terrains: Implications for the thermal structure of Archean and Proterozoic lithosphere. Journal of Geophysical Research 98: 12207–12218. Okrusch M, Schro¨der B, and Schnu¨tgen A (1979) Granulite-facies metabasite ejecta in the Laacher Sea area, Eifel, West Germany. Lithos 12: 251–270. O’Reilly SY, Griffin WL, and Stabel A (1988) Evolution of Phanerozoic eastern Australian lithosphere: Isotopic evidence for magmatic and tectonic underplating. In: Menzies MA and Cox KG (eds.) Oceanic and Continental Lithosphere: Similarities and Differences, Special Volume Journal of Petrology, pp. 89–108. Oxford: Oxford University Press. Owen JV, Greenough JD, Hy C, and Ruffman A (1988) Xenoliths in a mafic dyke at Popes Harbour, Nova Scotia: Implications for the basement to the Meguma Group. Canadian Journal of Earth Sciences 25: 1464–1471. Padovani ER and Carter JL (1977) Aspects of the deep crustal evolution beneath south central New Mexico. In: Heacock JG (ed.) The Earth’s Crust, pp. 19–55. Washington, DC: American Geophysical Union. Padovani ER and Hart SR (1981) Geochemical constraints on the evolution of the lower crust beneath the Rio Grande rift. Conference on the Processes of Planetary Rifting, pp. 149–152. Lunar and Planetary Science Institute. Padovani ER, Hall J, and Simmons G (1982) Constraints on crustal hydration below the Colorado Plateau from Vp measurements on crustal xenoliths. Tectonophysics 84: 313–328. Patchett PJ and Samson SD (2003) Ages and growth of the continental crust from radiogenic isotopes. In: Holland HD and Turekian KK (eds.) Treatise on Geochemistry, vol 3, pp. 321–348. Oxford: Elsevier. Pearcy LG, DeBari SM, and Sleep NH (1990) Mass balance calculations for two sections of island arc crust and implications for the formation of continents. Earth and Planetary Science Letters 96: 427–442. Pearson NJ, O’Reilly SY, and Griffin WL (1995) The crust-mantle boundary beneath cratons and craton margins: A transect across the south-west margin of the Kaapvaal craton. Lithos 36(3–4): 257–287. Percival JA and Card KD (1983) Archean crust as revealed in the Kapuskasing uplift, Superior province, Canada. Geology 11: 323–326. Percival JA, Fountain DM, and Salisbury MH (1992) Exposed cross sections as windows on the lower crust. In: Fountain DM, Arculus R, and Kay RW (eds.) Continental Lower Crust, pp. 317–362. Amsterdam: Elsevier. Peucker-Ehrenbrink B and Jahn B-M (2001) Rhenium-osmium isotope systematics and platinum group element concentations: Loess and the upper continental crust. Geochemistry, Geophysics, Geosystems 2: 1061. Phillips RJ, Raubertas RF, and Arvidson RE, et al. (1992) Impact craters and Venus resurfacing history. Journal of Geophysical Research 97(E10): 15923–15948. Pinet C and Jaupart C (1987) The vertical distribution of radiogenic heat production in the Precambrian crust of Norway and Sweden: Geothermal implications. Geophysical Research Letters 14: 260–263. Plank T and Langmuir CH (1998) The chemical composition of subducting sediment and its consequences for the crust and mantle. Chemical Geology 145: 325–394. Poldervaart A (1955) The chemistry of the Earth’s crust. Geological Society of America Special Paper 62: 119–144.

50

Composition of the Continental Crust

Quick JE, Sinigoi S, and Mayer A (1995) Emplacement of mantle peridotite in the lower continental crust, Ivrea-Verbano zone, northwest Italy. Geology 23(8): 739–742. Reid MR, Hart SR, Padovani ER, and Wandless GA (1989) Contribution of metapelitic sediments to the composition, heat production, and seismic velocity of the lower crust of southern New Mexico. Earth and Planetary Science Letters 95: 367–381. Roberts S and Ruiz J (1989) Geochemical zonation and evolution of the lower crust in Mexico. Journal of Geophysical Research 94: 7961–7974. Rogers NW (1977) Granulite xenoliths from Lesotho kimberlites and the lower continental crust. Nature 270: 681–684. Rogers NW and Hawkesworth CJ (1982) Proterozoic age and cumulate origin for granulite xenoliths, Lesotho. Nature 299: 409–413. Ronov AB and Yaroshevsky AA (1967) Chemical structure of the Earth’s crust. Geokhimiya 11: 1285–1309. Ronov AB and Yaroshevsky AA (1976) A new model for the chemical structure of the Earth’s crust. Geokhimiya 12: 1761–1795. Ross DC (1985) Mafic gneissic complex (batholithic root?) in the southernmost Sierra Nevada, California. Geology 13: 288–291. Rudnick RL (1990a) Continental crust: Growing from below. Nature 347: 711–712. Rudnick RL (1990b) Nd and Sr isotopic compositions of lower crustal xenoliths from North Queensland, Australia: Implications for Nd model ages and crustal growth processes. Chemical Geology 83: 195–208. Rudnick RL (1992) Xenoliths—samples of the lower continental crust. In: Fountain DM, Arculus R, and Kay RW (eds.) Continental Lower Crust, pp. 269–316. Amsterdam: Elsevier. Rudnick RL (1995) Making continental crust. Nature 378: 571–578. Rudnick RL and Cameron KL (1991) Age diversity of the deep crust in northern Mexico. Geology 19: 1197–1200. Rudnick RL and Goldstein SL (1990) The Pb isotopic compositions of lower crustal xenoliths and the evolution of lower crustal Pb. Earth and Planetary Science Letters 98: 192–207. Rudnick RL and Fountain DM (1995) Nature and composition of the continental crust: A lower crustal perspective. Reviews of Geophysics 33(3): 267–309. Rudnick RL and Presper T (1990) Geochemistry of intermediate to high-pressure granulites. In: Vielzeuf D and Vidal P (eds.) Granulites and Crustal Evolution, pp. 523–550. Amsterdam: Kluwer. Rudnick RL and Taylor SR (1987) The composition and petrogenesis of the lower crust: A xenolith study. Journal of Geophysical Research 92(B13): 13981–14005. Rudnick RL and Taylor SR (1991) Petrology and geochemistry of lower crustal xenoliths from northern Queensland and inferences on lower crustal composition. In: Drummond B (ed.) The Eastern Australian Lithosphere, Geological Society of Australia Special Publication, pp.189–208. Rudnick RL and Williams IS (1987) Dating the lower crust by ion microprobe. Earth and Planetary Science Letters 85: 145–161. Rudnick RL, McLennan SM, and Taylor SR (1985) Large ion lithophile elements in rocks from high-pressure granulite facies terrains. Geochimica et Cosmochimica Acta 49: 1645–1655. Rudnick RL, McDonough WF, McCulloch MT, and Taylor SR (1986) Lower crustal xenoliths from Queensland, Australia: Evidence for deep crustal assimilation and fractionation of continental basalts. Geochimica et Cosmochimica Acta 50: 1099–1115. Rudnick RL, McDonough WF, and O’Connell RJ (1998) Thermal structure, thickness and composition of continental lithosphere. Chemical Geology 145: 399–415. Rudnick RL, Ireland TR, Gehrels G, Irving AJ, Chesley JT, and Hanchar JM (1999) Dating mantle metasomatism: U–Pb geochronology of zircons in cratonic mantle xenoliths from Montana and Tanzania. In: Gurney JJ, Gurney JL, Pascoe MD, and Richardson SR (eds.) Proceedings of the VIIth International Kimberlite Conference, pp. 728–735. Cape Town: Red Roof Design. Rudnick RL, Barth M, Horn I, and McDonough WF (2000) Rutile-bearing refractory eclogites: Missing link between continents and depleted mantle. Science 287: 278–281. Ruiz J, Patchett PJ, and Arculus RJ (1988a) Nd–Sr isotope composition of lower crustal xenoliths—evidence for the origin of mid-Tertiary felsic volcanics in Mexico. Contributions to Mineralogy and Petrology 99: 36–43. Ruiz J, Patchett PJ, and Ortega-Gutierrez F (1988b) Proterozoic and Phanerozoic basement terranes of Mexico from Nd isotopic studies. Geological Society of America Bulletin 100: 274–281. Rutter MJ (1987) The nature of the lithosphere beneath the Sardinian continental block: Mantle and deep crustal inclusions in mafic alkaline lavas. Lithos 20: 225–234. Saal AE, Rudnick RL, Ravizza GE, and Hart SR (1998) Re–Os isotope evidence for the composition, formation and age of the lower continental crust. Nature 393: 58–61. Sachs PM and Hansteen TH (2000) Pleistocene underplating and metasomatism of the lower continental crust: A xenolith study. Journal of Petrology 41(3): 331–356.

Saleeby JB (1990) Progress in tectonic and petrogenetic studies in an exposed cross-section of young (c. 100 Ma) continental crust southern Sierra Nevada, California. In: Salisbury MH and Fountain DM (eds.) Exposed Cross-Sections of the Continental Crust, pp. 137–159. Norwell, MA: Kluwer Academic. Schaaf P, Heinrich W, and Besch T (1994) Composition and Sm–Nd isotopic data of the lower crust beneath San-Luis-Potosi, Central Mexico—evidence from a granulite-facies xenolith suite. Chemical Geology 118(1–4): 63–84. Schaber GG, Strom RG, Moore HJ, et al. (1992) Geology and distribution of impact craters on Venus–what are they telling us. Journal of Geophysical Research 97(E8): 13257–13301. Scherer EK, Cameron KL, Johnson CM, Beard BL, Barovich KM, and Collerson KD (1997) Lu–Hf geochronology applied to dating Cenozoic events affecting lower crustal xenoliths from Kilbourne Hole, New Mexico. Chemical Geology 142: 63–78. Schmitz MD and Bowring SA (2000) The significance of U–Pb zircon dates in lower crustal xenoliths from the southwestern margin of the Kaapvaal craton, southern Africa. Chemical Geology 172: 59–76. Schmitz MD and Bowring SA (2003a) Constraints on the thermal evolution of continental lithosphere from U–Pb accessory mineral thermochronometry of lower crustal xenoliths, southern Africa. Contributions to Mineralogy and Petrology 144: 592–618. Schmitz MD and Bowring SA (2003b) Ultrahigh-temperature metamorphism in the lower crust during Neoarchean Ventersdorp rifting and magmatism, Kaapvaal craton, southern Africa. Geological Society of America Bulletin 115: 533–548. Schulze DJ and Helmstaedt H (1979) Garnet pyroxenite and eclogite xenoliths from the Sullivan Buttes latite, Chino valley, Arizona. In: Boyd FR and Meyer HOA (eds.) The Mantle Sample: Inclusions in Kimberlites and Other Volcanics, pp. 318–329. Washington, DC: American Geophysics Union. Sclater JG, Jaupart CJ, and Galson D (1980) The heat flow through oceanic and continental crust and the heat loss of the earth. Reviews of Geophysics and Space Physics 18: 269–311. Selverstone J and Stern CR (1983) Petrochemistry and recrystallization history of granulite xenoliths from the Pali-Aike volcanic field, Chile. American Mineralogist 68: 1102–1111. Sewell RJ, Hobden BJ, and Weaver SD (1993) Mafic and ultramafic mantle and deep-crustal xenoliths from Banks Peninsula, South-Island, New-Zealand. New Zealand Journal of Geology and Geophysics 36(2): 223–231. Shatsky V, Rudnick RL, and Jagoutz E (1990) Mafic granulites from Udachnaya pipe, Yakutia: Samples of Archean lower crust? Deep Seated Magmatism and Evolution of Lithosphere of the Siberian Platform, 23–24. Shatsky VS, Sobolev NV, and Pavlyuchenko VS (1983) Fassaı¨te-garnet-anorthite xenolith from the Udachnaya kimberlite pipe, Yakutia. Doklady Akademii Nauk SSSR 272(1): 188–192. Shaw DM, Reilly GA, Muysson JR, Pattenden GE, and Campbell FE (1967) An estimate of the chemical composition of the Canadian Precambrian shield. Canadian Journal of Earth Sciences 4: 829–853. Shaw DM, Dostal J, and Keays RR (1976) Additional estimates of continental surface Precambrian shield composition in Canada. Geochimica et Cosmochimica Acta 40: 73–83. Shaw DM, Cramer JJ, Higgins MD, and Truscott MG (1986) Composition of the Canadian Precambrian shield and the continental crust of the Earth. In: Dawson JB, Carswell DA, Hall J, and Wedepohl KH (eds.) The Nature of the Lower Continental Crust, vol. 24, pp. 257–282. London: Geological Society of London. Shaw DM, Dickin AP, Li H, McNutt RH, Schwarcz HP, and Truscott MG (1994) Crustal geochemistry in the Wawa-Foleyet region, Ontario. Canadian Journal of Earth Sciences 31(7): 1104–1121. Sims KWW, Newsom HE, and Gladney ES (1990) Chemical fractionation during formation of the Earth’s core and continental crust: Clues from As, Sb, W, and Mo. In: Newsom HE, Jones JH, and Newson JH (eds.) Origin of the Earth, pp. 291–317. Oxford: Oxford University Press. Smith DE, Zuber MT, Solomon SC, et al. (1999) The global topography of Mars and implications for surface evolution. Science 284(5419): 1495–1503. Smith RD, Cameron KL, McDowell FW, Niemeyer S, and Sampson DE (1996) Generation of voluminous silicic magmas and formation of mid-Cenozoic crust beneath north-central Mexico: Evidence from ignimbrites, associated lavas, deep crustal granulites, and mantle pyroxenites. Contributions to Mineralogy and Petrology 123: 375–389. Smithson SB (1978) Modeling continental crust-structural and chemical constraints. Geophysical Research Letters 5(9): 749–752. Stolz AJ (1987) Fluid activity in the lower crust and upper mantle: Mineralogical evidence bearing on the origin of amphibole and scapolite in ultramafic and mafic granulite xenoliths. Mineralogical Magazine 51: 719–732. Stolz AJ and Davies GR (1989) Metasomatized lower crustal and upper mantle xenoliths from north Queensland: Chemical and isotopic evidence bearing on the

Composition of the Continental Crust

composition and source of the fluid phase. Geochimica et Cosmochimica Acta 53: 649–660. Stosch H-G and Lugmair GW (1984) Evolution of the lower continental crust: Granulite facies xenoliths from the Eifel, West Germany. Nature 311: 368–370. Stosch H-G, Ionov DA, Puchtel IS, Galer SJG, and Sharpouri A (1995) Lower crustal xenoliths from Mongolia and their bearing on the nature of the deep crust beneath cental Asia. Lithos 36: 227–242. Strom RG, Schaber GG, and Dawson DD (1994) The global resurfacing of Venus. Journal of Geophysical Research 99(E5): 10899–10926. Sun W, Bennett VC, Eggins SM, Kamenetsky VS, and Arculus RJ (2003) Evidence for enhanced mantle to crust rhenium transfer from undegassed arc magmas. Nature 422: 294–297. Sutherland FL and Hollis JD (1982) Mantle-lower crust petrology from inclusions in basaltic rocks in eastern Australia—an outline. Journal of Volcanology and Geothermal Research 14: 1–29. Tanaka T and Aoki K-I (1981) Petrogenetic implications of REE and Ba data on mafic and ultramafic inclusions from Itinome-gata, Japan. Journal of Geology 89: 369–390. Taylor SR (1964) Abundance of chemical elements in the continental crust—a new table. Geochimica et Cosmochimica Acta 28: 1273–1285. Taylor SR (1967) The origin and growth of continents. Tectonophysics 4: 17–34. Taylor SR (1977) Island arc models and the composition of the continental crust. In: Talwani M (ed.) Island Arcs, Deep Sea Trenches and Back-Arc Basins, pp. 325–336. Washington, DC: American Geophysical Union. Taylor SR (1992) Solar System Evolution. Cambridge: Cambridge University Press. Taylor SR and McLennan SM (1981) The composition and evolution of the continental crust: Rare Earth element evidence from sedimentary rocks. Philosophical Transactions of the Royal Society Series A 301: 381–399. Taylor SR and McLennan SM (1985) The Continental Crust: Its Composition and Evolution. Oxford: Blackwell. Taylor SR and McLennan SM (1995) The geochemical evolution of the continental crust. Reviews of Geophysics 33: 241–265. Taylor SR, McLennan SM, and McCulloch MT (1983) Geochemistry of loess, continental crustal composition and crustal model ages. Geochimica et Cosmochimica Acta 47: 1897–1905. Teng F, McDonough WF, Rudnick RL, et al. (2004) Lithium isotopic composition and concentration of the upper continental crust. Geochimica et Cosmochimica Acta 68: 4167–4178. Thomas CW and Nixon PH (1987) Lower crustal granulite xenoliths in carbonatite volcanoes of the western rift of East Africa. Mineralogical Magazine 51: 621–633. Toft PB, Hills DV, and Haggerty SE (1989) Crustal evolution and the granulite to eclogite transition in xenoliths from kimberlites in the West African craton. Tectonophysics 161: 213–231. Tredoux M, Hart RJ, Carlson RW, and Shirey SB (1999) Ultramafic rocks at the center of the Vredefort structure: Further evidence for the crust on edge model. Geology 27(10): 923–926. Trzcienski WE and Marchildon N (1989) Kyanite-garnet-bearing Cambrian rocks and Grenville granulites from the Ayers Cliff, Quebec, Canada, Lamprophyre Dike Suite—deep crustal fragments from the northern appalachians. Geology 17(7): 637–640. Upton BGJ, Aspen P, and Chapman NA (1983) The upper mantle and deep crust beneath the British Isles: Evidence from inclusion suites in volcanic rocks. Journal of the Geological Society 140: 105–122. Upton BGJ, Aspen P, Rex DC, Melcher F, and Kinny P (1998) Lower crustal and possible shallow mantle samples from beneath the Hebrides: Evidence from a xenolithic dyke at Gribun, western Mull. Journal of the Geological Society 155: 813–828. Upton BGJ, Aspen P, and Hinton RW (2001) Pyroxenite and granulite xenoliths from beneath the Scottish northern Highlands terrane: Evidence for lower-crust/uppermantle relationships. Contributions to Mineralogy and Petrology 142(2): 178–197. Urrutia-Fucugauchi J and Uribe-Cifuentes RM (1999) Lower-crustal xenoliths from the Valle de Santiago maar field, Michoacan-Guanajuato volcanic field, central Mexico. International Geology Review 41(12): 1067–1081. van Breeman O and Hawkesworth CJ (1980) Sm–Nd isotopic study of garnets and their metamorphic host rocks. Transactions of the Royal Society of Edinburgh 71: 97–102. van Calsteren PWC, Harris NBW, Hawkesworth CJ, Menzies MA, and Rogers NW (1986) Xenoliths from southern Africa: A perspective on the lower crust. In: Dawson JB, Carswell DA, Hall J, and Wedepohl KH (eds.) The Nature of the Lower Continental Crust, Geological Society Special Publication, vol. 25, pp. 351–362. London, Geological Society of London. Vidal P and Postaire B (1985) E´tude par la me´thode Pb–Pb de roches de haut grade me´tamorphique implique´es dans la chaıˆne Hercynienne. Chemical Geology 49: 429–449.

51

Vielzeuf D (1983) The spinel and quartz associations in high grade xenoliths from Tallante (S. E. Spain) and their potential use in geothermometry and barometry. Contributions to Mineralogy and Petrology 82: 301–311. Villaseca C, Downes H, Pin C, and Barbero L (1999) Nature and composition of the lower continental crust in central Spain and the granulite-granite linkage: Inferences from granulitic xenoliths. Journal of Petrology 40(10): 1465–1496. Voshage H, Hofmann AW, and Mazzucchelli M, et al. (1990) Isotopic evidence from the Ivrea zone for hybrid lower crust formed by magmatic underplating. Nature 347: 731–736. Wanke H, Bruckner J, Dreibus G, Rieder R, and Ryabchikov I (2001) Chemical composition of rocks and soils at the pathfinder site. Space Science Reviews 96(1–4): 317–330. Warren RG, Kudo AM, and Keil K (1979) Geochemistry of lithic and single-crystal inclusions in basalts and a characterization of the upper mantle-lower crust in the Engle Basin, Rio Grande Rift, New Mexico. In: Riecker RE (ed.) Rio Grande Rift: Tectonics and Magmatism, pp. 393–415. Washington, DC: American Geophysical Union. Wass SY and Hollis JD (1983) Crustal growth in southeastern Australia—evidence from lower crustal eclogitic and granulitic xenoliths. Journal of Metamorphic Geology 1: 25–45. Weaver BL and Tarney J (1980) Continental crust composition and nature of the lower crust: Constraints from mantle Nd–Sr isotope correlation. Nature 286: 342–346. Weaver BL and Tarney J (1981) Lewisian gneiss geochemistry and Archaean crustal development models. Earth and Planetary Science Letters 55: 171–180. Weaver BL and Tarney J (1984) Empirical approach to estimating the composition of the continental crust. Nature 310: 575–577. Weber MBI, Tarney J, Kempton PD, and Kent RW (2002) Crustal make-up of the northern Andes: Evidence based on deep crustal xenolith suites, Mercaderes, SW Colombia. Tectonophysics 345(1–4): 49–82. Wedepohl H (1995) The composition of the continental crust. Geochimica et Cosmochimica Acta 59: 1217–1239. Wedepohl KH (1969–1978) Handbook of Geochemistry. Berlin: Springer. Wendlandt E, DePaolo DJ, and Baldridge WS (1993) Nd and Sr isotope chronostratigraphy of Colorado Plateau lithosphere: Implications for magmatic and tectonic underplating of the continental crust. Earth and Planetary Science Letters 116: 23–43. Wendlandt E, DePaolo DJ, and Baldridge WS (1996) Thermal history of Colorado Plateau lithosphere from Sm–Nd mineral geochronology of xenoliths. Geological Society of America Bulletin 108(7): 757–767. Wernicke B, Clayton R, Ducea M, et al. (1996) Origin of high mountains in the continents: The southern Sierra Nevada. Science 271: 190–193. Wilde SA, Valley JW, Peck WH, and Graham CM (2001) Evidence from detrital zircons for the existence of continental crust and oceans on the Earth 4.4 Gyr ago. Nature 409(6817): 175–178. Wilkinson JFG (1975) An Al-spinel ultramafic-mafic inclusion suite and high pressure megacrysts in an analcimite and their bearing on basaltic magma fractionation at elevated pressures. Contributions to Mineralogy and Petrology 53: 71–104. Wilkinson JFG and Taylor SR (1980) Trace element fractionation trends of thoeiitic magma at moderate pressure: Evidence from an Al-spinel ultramafic-mafic inclusion suite. Contributions to Mineralogy and Petrology 75: 225–233. Wilshire HW, Meyer CE, and Nakata JK, et al. (1988) Mafic and Ultramafic Xenoliths from Volcanic Rocks of the Western United States US Geological Survey Professional Paper 1443. Washington, DC: US Government Printing Office. Windley BF (1995) The Evolving Continents. Wiley. Wo¨rner G, Schmincke H-U, and Schreyer W (1982) Crustal xenoliths from the Quaternary Wehr volcano (East Eifel). Neues Jahrbuch fu¨r Mineralogie (Abhandlungen) 144(1): 29–55. Yardley BWD (1986) Is there water in the deep continental crust? Nature 323: 111. Yu JH, O’Reilly SY, Griffin WL, Xu XS, Zhang M, and Zhou XM (2003) The thermal state and composition of the lithospheric mantle beneath the Leizhou Peninsula South China. Journal of Volcanology and Geothermal Research 122(3–4): 165–189. Zashu S, Kaneoka I, and Aoki K-I (1980) Sr isotope study of mafic and ultramafic inclusions from Itinome-gata, Japan. Geochemical Journal 14: 123–128. Zheng JP, Sun M, Lu FX, and Pearson N (2003) Mesozoic lower crustal xenoliths and their significance in lithospheric evolution beneath the Sino-Korean craton. Tectonophysics 361(1–2): 37–60. Zhou XH, Sun M, Zhang GH, and Chen SH (2002) Continental crust and lithospheric mantle interaction beneath North China: Isotopic evidence from granulite xenoliths in Hannuoba, Sino-Korean craton. Lithos 62(3–4): 111–124.